US5878172A - Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform - Google Patents

Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform Download PDF

Info

Publication number
US5878172A
US5878172A US09/069,870 US6987098A US5878172A US 5878172 A US5878172 A US 5878172A US 6987098 A US6987098 A US 6987098A US 5878172 A US5878172 A US 5878172A
Authority
US
United States
Prior art keywords
edge
image
layer
images
pixels
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US09/069,870
Inventor
Shiyu Go
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Inphi Corp
Original Assignee
Oki Electric Industry Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from JP6264836A external-priority patent/JPH08123967A/en
Priority claimed from JP15014295A external-priority patent/JP3308767B2/en
Application filed by Oki Electric Industry Co Ltd filed Critical Oki Electric Industry Co Ltd
Priority to US09/069,870 priority Critical patent/US5878172A/en
Application granted granted Critical
Publication of US5878172A publication Critical patent/US5878172A/en
Assigned to OKI ELECTRIC INDUSTRY CO., LTD. reassignment OKI ELECTRIC INDUSTRY CO., LTD. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: GO, SHIYU
Assigned to GLOBAL D, LLC. reassignment GLOBAL D, LLC. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: OKI ELECTRIC INDUSTRY CO., LTD.
Assigned to INPHI CORPORATION reassignment INPHI CORPORATION ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: GLOBAL D, LLC.
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/85Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51Motion estimation or motion compensation
    • H04N19/537Motion estimation other than block-based
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H04N19/63Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/30Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability
    • H04N19/39Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability involving multiple description coding [MDC], i.e. with separate layers being structured as independently decodable descriptions of input picture data

Definitions

  • This invention relates to a compressive image encoding and decoding method using edge synthesis and the inverse wavelet transform, and to digital image encoding and decoding devices employing this method.
  • Compression is essential for efficient storage and transmission of digitized images. Compression methods have been described by the Joint Photographic Experts Group (JPEG) for still images, and the Motion Picture Experts Group (MPEG) for moving images.
  • JPEG Joint Photographic Experts Group
  • MPEG Motion Picture Experts Group
  • the JPEG method involves a discrete cosine transform (DCT), followed by quantization and variable-length encoding.
  • DCT discrete cosine transform
  • the MPEG method involves detecting motion vectors. Both methods require substantial computation, the detection of motion vectors being particularly demanding.
  • This transform employs a family of wavelets related by dilation and translation; that is, the family consists of occurrences of the same basic wavelet at different locations and on different scales. If the scales form a progressively doubling sequence, and if the basic wavelet is zero everywhere except in a limited domain, wavelet transforms and inverse wavelet transforms can be carried out with efficient computational algorithms.
  • a wavelet transform can be described as a filtering process executed at each wavelet scale.
  • a digitized image for example, is transformed by filtering with the basic wavelet, then with the basic wavelet dilated by a factor of two, then with the basic wavelet dilated by a factor of four, and so on.
  • One prior-art wavelet encoding scheme employs a complementary pair of wavelets to divide an image into a high-frequency component and a low-frequency component. These components contain information about variations on scales respectively less than and greater than a certain cut-off scale. This process is iterated on the low-frequency component with a doubling of the wavelet scale, obtaining new low-frequency and high-frequency components, then iterated again on the new low-frequency component, and so on. After a certain number of iterations, the components are encoded by an encoding scheme that works from low-toward high-frequency information. This scheme enables accurate image reconstruction, but retains too much high-frequency information to achieve high compression ratios.
  • Another prior-art wavelet encoding scheme employs a basic wavelet that is the first derivative of a smoothing filter (that is, the first derivative of a low-pass filtering function).
  • This type of wavelet acts as a high-pass filter.
  • High-frequency information is obtained by detecting local peaks (local maxima of absolute values) in the result of the wavelet transform, which correspond to edges in the original image.
  • the size and location of the peak values at a selected scale are encoded, along with a low-frequency image obtained by smoothing at the largest scale of the wavelet transform. Fairly high compression ratios can be obtained in this way.
  • this prior-art method employs an algorithm derived from a mathematical procedure involving iterated projections in Hilbert space. Under ideal conditions, the projections converge toward a unique set of data that (i) have the required local peak values and (ii) are within the range of the wavelet transform operator. An inverse wavelet transform is then carried out on the converged data to obtain the original image.
  • edge-based encoding schemes in general is that encoding efficiency tends to be impaired by edge fragmentation and irregularity.
  • Another object of the invention is to encode moving images efficiently.
  • Still another object is to reduce irregularity of edges.
  • Yet another object is to avoid fragmentation of edges.
  • Still another object is to reconstruct digitized images from encoded information efficiently and reliably.
  • Yet another object is to reconstruct digitized moving images efficiently and reliably.
  • the invented method of encoding a digitized image comprises the steps of:
  • edge image information encoding the position and sharpness of the detected edges to generate edge image information
  • Detecting edges in the digitized image may comprise the further steps of thickening the detected edges, then trimming each edge to a uniform thickness.
  • the invented method of reconstructing a digitized image encoded by the invented encoding method comprises the steps of:
  • FIG. 1 is a block diagram of the digital image encoder in a first embodiment of the invention.
  • FIG. 2 is a block diagram of the digital image decoder in the first embodiment.
  • FIG. 3 is a graph illustrating pixel values at an edge.
  • FIG. 4 is a block diagram of the digital image encoder in a second embodiment of the invention.
  • FIG. 5 is flow diagram illustrating the smoothing and down-sampling process in the encoder of the second embodiment.
  • FIG. 6 is a block diagram of the digital image decoder in the second embodiment.
  • FIG. 7 is a flow diagram illustrating the synthesis of high-frequency images in the second embodiment.
  • FIG. 8 is a flow diagram illustrating the inverse wavelet transform in the second embodiment.
  • FIG. 9 is a block diagram of the digital image encoder in a third embodiment of the invention.
  • FIG. 10 is a flow diagram illustrating one-dimensional filtering and down-sampling in the third embodiment.
  • FIG. 11 is a block diagram of the digital image decoder in the third embodiment.
  • FIG. 12 is a flow diagram illustrating the synthesis of high-frequency images in the third embodiment.
  • FIG. 13 is a flow diagram illustrating the inverse wavelet transform in the third embodiment.
  • FIG. 14 is a diagram of an edge, illustrating irregular shape and fragmentation.
  • FIG. 15 illustrates thickening of the edge in FIG. 14 according to a fourth embodiment of the invention.
  • FIG. 16 illustrates trimming of the edge in FIG. 15 according to the fourth embodiment.
  • FIG. 17 is a block diagram of the digital image encoder in a fifth embodiment, for encoding moving images.
  • FIG. 18 is a block diagram of the digital image decoder in the fifth embodiment.
  • FIG. 19 is a block diagram of the digital image encoder in a sixth embodiment of the invention.
  • FIG. 20 is a block diagram of the digital image encoder in a seventh embodiment.
  • FIG. 21 is a block diagram of the digital image decoder in the seventh embodiment.
  • FIG. 22 is a block diagram of the digital image encoder in an eighth embodiment.
  • FIG. 23 is a block diagram of the digital image encoder in a ninth embodiment.
  • FIG. 24 is a block diagram of the digital image decoder in the ninth embodiment.
  • FIG. 25 is a block diagram of the digital image encoder in a tenth embodiment.
  • FIG. 26 is a block diagram of the digital image decoder in the tenth embodiment.
  • FIG. 27 is a block diagram of a digital image encoder/decoder in an eleventh embodiment of the invention.
  • FIG. 28 is a block diagram of a digital image encoder/decoder in a twelfth embodiment.
  • a (two-dimensional) digitized image is an array of pixels having values x(i, j), where the horizontal coordinate i and vertical coordinate j range over sets of integers. The range of these coordinates is the size of the image, i.e., the number of pixels in the horizontal and vertical directions.
  • the pixel values represent, for example, intensity gradations.
  • a one-dimensional image is a horizontal or vertical line in a two-dimensional digitized image, i.e., a set of pixels with values x(i) indexed by a single coordinate.
  • a standard edge is a one-dimensional image having just one sharp variation point.
  • An example would be an image with pixel values that change first at one constant rate, then at a different constant rate, such as the following:
  • a standard edge serves as a basic model for all the edges occurring in a digitized image.
  • Down-sampling a digitized image means reducing its size by a process generally analogous to photographic reduction: by selecting every n-th pixel, for example, where n is a suitable integer. Up-sampling means increasing the size of an image by interpolating new pixels between the existing pixels.
  • a (one-dimensional) filter is a set of coefficients f(k), where k ranges over the integers.
  • the non-zero coefficients f(k) are the taps of the filter.
  • the Fourier transform F of a filter f is defined as follows, where summation is over all taps, e is the natural logarithm base, and j is the square root of minus one. ##EQU1##
  • the conjugate f* of a filter f is obtained by reversing the order of coefficients:
  • the Fourier transform of f* is therefore the complex conjugate of the Fourier transform of f.
  • a filter has even symmetry around i 0 if it has equal values on both sides of i 0 ; that is, for all values of k:
  • a filter has even symmetry around zero if it is equal to its own conjugate.
  • a filter has odd symmetry around i 0 if it has opposite values on both sides of i 0 , that is:
  • a filter satisfies the exact reconstruction condition if its Fourier transform F satisfies the following condition for all values of ⁇ :
  • Filters satisfying these exact reconstruction conditions are well known in the wavelet transform art.
  • a pair of filters satisfying the second exact reconstruction condition above is referred to as a complementary pair.
  • Filtering can be done two-dimensionally by, for example, filtering first in the horizontal direction, then in the vertical direction, or vice versa.
  • a high-pass filter is a filter that retains small-scale variations and rejects large-scale variations.
  • a low-pass filter rejects small-scale variations and retains large-scale variations.
  • Smoothing filter is a synonym for low-pass filter.
  • the cut-off frequency of a high-pass or low-pass filter indicates the scale that divides the retained variations from the rejected variations.
  • a lower cut-off frequency corresponds to a larger scale of variations.
  • Frequency actually refers to the variable ⁇ in the Fourier transform F( ⁇ ) of the filter.
  • the digital image encoder in the first embodiment comprises an input terminal 10, a two-dimensional smoothing and down-sampling section 20, and an edge detector 21.
  • the digital image encoder also comprises an edge image encoder 24, a reduced image encoder 25, and a multiplexer 26, which together constitute an encoding section 27.
  • a digitized image X 0 is input from the input terminal 10 to the edge detector 21 and first smoothing filter 22-1.
  • Smoothing filter 22-1 outputs a filtered image X 1 to the next smoothing filter 22-2, and this process continues, each smoothing filter 22-m filtering the image X m-1 received from the preceding smoothing filter and furnishing a new filtered image X m to the next smoothing filter, until the last smoothing filter 22-n is reached.
  • the last smoothing filter 22-n outputs a low-frequency image X n to the down sampler 23, which outputs a reduced image R to the reduced image encoder 25.
  • the reduced image encoder 25 encodes the reduced image R to obtain reduced image information Cr.
  • the edge detector 21 detects edge points in the input image X 0 and outputs horizontal and vertical edge images Sh and Sv to the edge image encoder 24 in the encoding section 22.
  • the edge image encoder 24 encodes these edge images to obtain edge image information Cs.
  • the multiplexer 26 combines the reduced image information Cr and edge image information Cs into an encoded image C, which is output to a first input/output device 28 such as a communication channel, semiconductor memory, or magnetic disk storage device.
  • the digital image decoder has a decoding section 29 comprising a demultiplexer 30, an edge image decoder 31, and a reduced image decoder 32.
  • the demultiplexer 30 receives the encoded image C from the first input/output device 28 and separates it into edge image information Cs, which it supplies to the edge image decoder 31, and reduced image information Cr, which it supplies to the reduced image decoder 32.
  • the reduced image decoder 32 decodes Cr to obtain a decoded reduced image R', which the up sampler 35 enlarges to obtain a decoded low-frequency image X' n .
  • each inverse wavelet processor 36-m receives the corresponding high-frequency images Yh m and Yv m and a partially reconstructed image X' m from the preceding inverse wavelet processor 36-(m+1) in the cascade, and outputs a partially reconstructed image X' m-1 to the next inverse wavelet processor 36-(m-1).
  • the first inverse wavelet processor 36-n in the cascade receives the decoded low-frequency image X' n from the up sampler 35.
  • the last inverse wavelet processor 36-1 outputs the fully reconstructed image X' 0 to a second input/output device 37.
  • Each of the elements in the FIGS. 1 and 2 comprises a memory for storing data and well-known circuits for performing arithmetic and logic operations. Descriptions of the circuit configurations will be omitted to avoid obscuring the invention with irrelevant detail.
  • the entire digital image encoder can be fabricated as a single semiconductor integrated circuit, or can be built from a small number of such circuits; likewise the digital image decoder.
  • the invention can also be practiced by providing a general-purpose processor such as a microprocessor or digital signal processor with memory for storing data and programs for executing the functions of the individual elements in FIG. 1 or FIG. 2.
  • the input image X 0 is a digitized image with pixel values X 0 (i, j).
  • the edge detector 21 detects edges by detecting variations in the rate of change of pixel values from pixel to pixel. More specifically, the edge detector 21 takes differences between adjacent pixel values, then takes differences between these differences, and finally performs a thresholding operation to reduce small differences to zero. This process is carried out separately in the horizontal and vertical directions.
  • the edge detector 21 computes the following pixel difference Pdh(i, j):
  • Equivalent edge detection could be performed by filtering the image X 0 with a second-derivative filter having coefficients (1, -2, 1), but taking differences of differences is preferable because it requires less computation.
  • the detected value Sh(i, j) is the edge sharpness at pixel (i, j), also referred to as the edge size.
  • Edge sharpness values with absolute values equal to or less than a certain threshold value T are quantized to zero. Values exceeding T are also quantized, to reduce the number of bits in the edge-image data; then the quantized sharpness values Sh(i, j) at all pixels are output as the horizontal edge image Sh.
  • >T, are referred to as edge points.
  • the meaning of the edge sharpness computation can be seen in FIG. 3.
  • the horizontal axis represents, for example, the horizontal direction in the image, and the vertical axis represents the pixel value.
  • the difference between pixel values 38 and 39 is -A
  • the difference between pixel values 40 and 38 is B
  • the other pixels in FIG. 3 have zero sharpness.
  • FIG. 3 thus illustrates a standard edge, with just one sharp variation.
  • the vertical edge image is generated in the same way, by taking differences in the vertical direction, then taking differences of these differences.
  • the edge sharpness Sv in the vertical direction is:
  • the vertical edge image is quantized as was the horizontal edge image, points for which
  • Quantization can be performed by dividing the edge sharpness values by a power of two, for example, or by using a quantization table.
  • Sh and Sv will be employed herein to denote edge sharpness values both before and after quantization.
  • Smoothing filter 22-1 employs a low-pass filter h 1 with even symmetry and a certain number of taps.
  • this filter either has 2N+1 taps and is symmetric around zero, or has 2N taps and is symmetric around 1/2, where N is an appropriate positive integer.
  • the even symmetry condition in the first case is:
  • the low-pass filter h 1 satisfies the exact reconstruction condition, and the sum of its tap coefficients is plus or minus one. That is, the Fourier transform H 1 of h 1 satisfies the following conditions:
  • the smoothing filter 22-1 performs a two-dimensional low-pass filtering operation by filtering the input image X 0 with h 1 horizontally and vertically, obtaining: ##EQU7##
  • the other smoothing filters 22-m also employ low-pass filters h m with even symmetry, and with tap coefficients summing to plus or minus one.
  • the cut-off frequency of each smoothing filter 22-m (m>1) is one-half the cut-off frequency of the preceding filter 22-(m-1) in the cascade. Filters with these cut-off frequencies can be obtained by doubling the tap spacing from one filter to the next, and inserting zeros to fill the vacated spaces. If the number of taps is odd, for example, then h m can be obtained from h m-1 as follows:
  • image X m-1 is filtered horizontally and vertically by filter h m to obtain image X m .
  • filter h m As the image passes through the cascade, variations on progressively larger scales are smoothed out, and resolution of fine detail decreases correspondingly.
  • the low-pass filters h m (m>1) are obtained by doubling the tap spacing as described above, they will pass high-frequency variations in certain regions above their cut-off frequencies. This is allowable because those high-frequency variations will already have been removed by previous filters in the cascade.
  • the down sampler 23 down-samples the low-frequency image X n output by the smoothing filter 22-n, thereby reducing the amount of image data.
  • the reduced image R output by the down sampler 23 resembles a photographic reduction of the original image X 0 .
  • the edge image encoder 24 encodes the quantized horizontal and vertical edge images Sh and Sv by using, for example, run-length encoding to encode the positions of edge points, and differential encoding to encode the sharpness values at these points.
  • chain encoding can be used to encode the edge positions, taking advantage of the fact that edge points tend to be organized into chains of mutually contiguous points. Chain encoding encodes only the displacement from one position to the next in such a chain.
  • the reduced image encoder 25 encodes the reduced image R by, for example, performing a discrete cosine transform, quantizing the resulting DCT coefficients, and encoding them in zig-zag order by a variable-length encoding method.
  • the reduced image R can be encoded by a predictive encoding method similar to the differential pulse-code modulation method (DPCM) commonly employed for audio signals, or by any other suitable method.
  • DPCM differential pulse-code modulation method
  • the encoded image C consisting of the edge image information Cs and low-frequency information Cr as multiplexed by the multiplexer 26, is output to the first input/output device 28 for transmission or storage.
  • the amount of encoded data is generally reduced to a small fraction of the amount in the original image X 0 , so the encoded image C can be efficiently transmitted or stored.
  • Filter g 1 is a high-pass filter related to the smoothing filter h 1 as follows:
  • These filters g m can be obtained from g 1 by successively doubling the tap spacing, in the same way that the filters h m were obtained from h 1 .
  • Filter f m is a completion filter, defined in terms of its Fourier transform F m as follows. ##EQU8##
  • F m is well known in the wavelet transform art and comes from the following equation:
  • Filter e m is an edge synthesis filter derived by filtering a standard edge of unit sharpness by the corresponding high-pass filter g m .
  • g m has even symmetry, all standard edges with unit sharpness lead to the same e m , which has the following tap values e m (t).
  • decoders 31 and 32 carry out processes that are reverse to the processes performed by encoders 24 and 25 in FIG. 1. For example, if the reduced image encoder 25 in FIG. 1 performed a discrete cosine transform followed by quantization and variable-length encoding, the reduced image decoder 32 in FIG. 2 performs variable-length decoding, dequantization, then an inverse discrete cosine transform to obtain the decoded reduced image R'.
  • the edge image decoder 31 performs run-length or chain decoding to recover the edge positions and an accumulation process to obtain the sharpness values.
  • the edge image decoder 31 also dequantizes the sharpness values.
  • the up sampler 35 obtains the decoded low-frequency image X' n by restoring the decoded reduced image R' to the original image size (the size of the edge images and high-frequency images). If, for example, each pixel value in the reduced image R represented an M ⁇ M block in the low-frequency image X n , the up sampler 35 can simply copy this value to all pixels in the corresponding M ⁇ M block of the decoded low-frequency image X' 1 . More sophisticated up-sampling methods involving interpolation can also be employed, but smooth interpolation is not required, because the image will be smoothed during the inverse wavelet transform.
  • the cascaded smoothing filters 33-m carry out successive two-dimensional filtering operations on the pairs of edge images Sh m and Sv m in the same way that the cascaded smoothing filters 22-m carried out successive filtering operations in the encoder. Specifically, Sh m is filtered by h m horizontally and vertically to obtain Sh m+1 , and Sv m is filtered by h m horizontally and vertically to obtain Sv m+1 .
  • Each edge synthesizer 34-m filters the corresponding horizontal edge image Sh m by the edge synthesis filter e m in the horizontal direction, and the vertical edge image Sv m by em in the vertical direction.
  • the smoothing filters h m have an odd number of taps and the high-pass filters gm accordingly have even symmetry
  • the resulting horizontal and vertical high-frequency images Yh m and Yv m can be described by the following equations: ##EQU10##
  • the edge synthesizer 34 synthesizes the result of filtering a standard edge of the corresponding sharpness by the high-frequency filter g m . It then adds these results together for all edge points to synthesize a high-frequency image.
  • the smoothing filters h m have an even number of taps and the high-pass filters gm accordingly have odd symmetry, it is advantageous to carry out the edge synthesis indirectly.
  • the results of filtering different standard edges by the same g m need not be the same.
  • the resulting edge synthesis filters em can have an unlimited number of non-zero taps.
  • the horizontal edge image Sh m is first filtered horizontally by de m (t) to obtain a horizontal differential high-frequency image ⁇ Yh, and the vertical edge image Sv m is filtered vertically with de(t) to obtain a vertical differential high-frequency image ⁇ Yv.
  • the horizontal and vertical differential high-frequency images are then cumulatively summed to obtain the horizontal and vertical high-frequency images Yh m and Yv m . If the horizontal and vertical coordinates i and j both start at zero, these accumulation operations can be described as follows:
  • each inverse wavelet processor 36-m performs three two-dimensional filtering operations, using the conjugate filters f* m , g* m , and h* m of f m , g m , and h m .
  • the image X' m is filtered by h* m horizontally and vertically
  • the horizontal high-frequency image Yh* m is filtered by g* m horizontally and f* m vertically
  • the vertical high-frequency image Yv m is filtered by g* m vertically and f* m horizontally.
  • the three results are added to obtain the partially reconstructed image X' m-1 .
  • This inverse wavelet transform is well known in the art, and is described by the following equation: ##EQU13##
  • the reconstructed image X' 0 is output to the second input/output device 37 for display or print-out.
  • the reconstructed image can be output to a computer for further image processing.
  • Synthesizing high-frequency image information from edge images as described above enables high compression ratios to be obtained.
  • High-frequency information corresponding to edges below the sharpness threshold T is lost, so there is some loss of low-level, high-frequency shading variations, but the reconstructed image is not distorted, and sharp edges are faithfully preserved.
  • edge detection is performed only at the original image scale, which permits edge detection to proceed concurrently with the smoothing filtering and down-sampling operations in the encoder, to speed up the encoding process.
  • edge detection is performed only at the original image scale, which permits edge detection to proceed concurrently with the smoothing filtering and down-sampling operations in the encoder, to speed up the encoding process.
  • high-frequency images are synthesized in a single step at each wavelet scale, without recourse to iterative procedures. This reduces the computational load in the decoder, speeds up the decoding process, and eliminates the prior-art problem of deciding when to stop iterating.
  • the second embodiment is similar to the first, but incorporates a down-sampling process into each of the smoothing filters in the encoder and decoder, and an up-sampling process into each of the inverse wavelet processors in the decoder.
  • FIG. 5 shows a conceptual flow diagram of the smoothing and down-sampling filtering process.
  • the first step 43 is a vertical smoothing step carried out using a low-pass filter h identical to filter h 1 in the first embodiment.
  • the second step 44 is a vertical down-sampling step that discards every other pixel in the vertical direction.
  • the third step 45 is a horizontal smoothing step in which the remaining image data are filtered by the same filter h in the horizontal direction.
  • the fourth step 46 is a horizontal down-sampling step that discards every other pixel in the horizontal direction.
  • This operation reduces the size of the image by a factor of two in both the horizontal and vertical directions.
  • hd denotes filtering by h followed by down-sampling as described above.
  • FIG. 6 shows the digital image decoder of the second embodiment.
  • the decoding section 29 is the same as in the first embodiment.
  • the output of the reduced image decoder 32 is now denoted X' n , but is similar to the output R in the first embodiment.
  • the decoder employs filters e, f, g, and h that are identical to e 1 , f 1 , g 1 , and h 1 in the first embodiment.
  • the subscripts d and u attached to these filters denote associated down-sampling and up-sampling operations, respectively.
  • Each filter 48-m filters and smooths each input edge image Sv m and Sh m in both the vertical and horizontal directions, and down-samples the result by a factor of two in both directions to generate edge images Sv m+1 and Sh m+1 . Due to the down-sampling, Sv m+1 is only half as wide and half as high as Sv m , and Sh m+1 is only half as wide and half as high as Sh m .
  • the vertical edge image Sv m is down-sampled by a factor of two in the horizontal direction, filtered by the above-mentioned filter e in the vertical direction, then down-sampled by a factor of two in the vertical direction
  • the horizontal edge image Sh m is down-sampled by a factor of two in the vertical direction, filtered by filter e in the horizontal direction, then down-sampled by a factor of two in the horizontal direction.
  • Processors 50-m use the conjugate filters f*, g*, and h* to perform the same computation as in the first embodiment, but each filtering step in the computation is preceded by an up-sampling step.
  • These up-sampling steps double the image size by inserting zero-valued pixels between every pair of adjacent pixels in the appropriate direction. To prevent the zero-valued pixels from reducing the mean value of the output image, after each filtering step, the resulting pixel values are doubled.
  • the image X' m-1 output from inverse wavelet processor 50-m is twice as wide and twice as high as each of the three images X' m , Yv m , and Yh m input to processor 50-m.
  • the three input images X' m , Yv m , and Yh m all have the same size.
  • FIG. 8 depicts the operation performed by the inverse wavelet processor 50-m schematically.
  • the partially reconstructed image X' m received from the preceding processor 50-(m+1) in the cascade is up-sampled horizontally by interpolating zeros (reducing the mean pixel value by a factor of two) and filtered horizontally by h*; then the resulting pixel values are doubled (restoring the correct mean value).
  • the image is now up-sampled vertically, again by interpolating zeros, and filtered by h* vertically.
  • high-frequency image Yh m is up-sampled vertically, filtered by f* vertically, doubled in pixel value, up-sampled horizontally, and filtered by g* horizontally.
  • High-frequency image Yv m is up-sampled horizontally, filtered by f* horizontally, doubled in pixel value, up-sampled vertically, and filtered by g* vertically.
  • the three resulting images are then added together to produce a sum image, all pixel values in which are again doubled to counteract the effect of the second up-sampling steps.
  • the final result is the partially reconstructed image X' m-1 which is passed to the next wavelet processor 50-(m-1), or the reconstructed image X' 0 output to the input/output device 37.
  • inverse wavelet processor 50-m The operation performed by inverse wavelet processor 50-m can be described by the following equation. Up-sampling is accomplished by dividing the values of i, j, k, and l by two. If either i/2+k/2 or j/2+1/2 is not an integer, the corresponding values of X' m , Yh m , and Yv m are taken to be zero. ##EQU15##
  • the second embodiment provides the same advantages as the first embodiment but requires less computation, because the amount of image data to be processed is reduced by a factor of four (by a factor of two horizontally and another factor of two vertically) at each wavelet scale.
  • the third embodiment detects edges diagonally as well as horizontally and vertically. To prevent an increase in the amount of encoded data, some of the down-sampling steps that were performed in the decoder in the second embodiment are moved into the encoder.
  • FIG. 9 is a block diagram of the digital image encoder of the third embodiment.
  • the two-dimensional smoothing and down-sampling section 41 is the same as in the second embodiment.
  • the original image X 0 is also input to a one-dimensional smoothing and down-sampling section 51, which outputs the original image X 0 together with a vertically smoothed and down-sampled image Xh and a horizontally smoothed and down-sampled image Xv to an edge detector 52.
  • the edge detector 52 outputs three edge images Sh, Sv, and Sc to the encoding section 27, which is the same as in the first two embodiments.
  • FIG. 10 illustrates the processes carried out in the one-dimensional smoothing and down-sampling section 51.
  • Image Xv is produced by filtering the original image X 0 with the low-pass filter h in the horizontal direction, then down-sampling the resulting image by removing every other pixel in the horizontal direction.
  • Image Xv accordingly contains only half as much data as image X 0 .
  • image Xh is produced by filtering X 0 vertically with the low-pass filter h, then down-sampling the result by removing every other pixel in the vertical direction.
  • Image Xh also contains only half as much data as image X 0 .
  • the edge detector 52 detects edges by the method described earlier: by taking differences between adjacent pixels, then taking differences between these differences. This method is applied in the directions that have not been down-sampled. Accordingly, differences are taken horizontally in the vertically down-sampled image Xh, producing a horizontal edge image Sh. Similarly, differences are taken vertically in the horizontally down-sampled image Xv, producing a vertical edge image Sv.
  • Edges are detected diagonally by processing the original image X 0 in both the horizontal and vertical directions.
  • the difference-of-differences operation is first performed on X 0 in, for example, the horizontal direction, producing an intermediate image.
  • the difference-of-differences operation is next performed on this intermediate image in the vertical direction.
  • Resulting pixels with absolute values equal to or less than a threshold value T are then set to zero, and other values are suitably quantized, producing a diagonal edge image Sc.
  • Sc Prior to quantization, Sc has the following values: ##EQU17##
  • the edge image encoder 24 can encode the diagonal edge image Sc by methods similar to those used for the vertical and horizontal edge images Sv and Sh. Sc has the same size as the original image X 0 , while Sv and Sh are only half this size, so the total amount of data to be encoded is the same as in the first two embodiments.
  • FIG. 11 is a block diagram of the digital image decoder in the third embodiment.
  • the decoding section 29 is the same as in the preceding embodiments, except that the edge image decoder 31 outputs a set of three edge images, Sv 1 , Sh 1 , and Sc 1 , corresponding to the encoded edge images Sv, Sh, and Sc.
  • Edge synthesizer 54-m operates as illustrated in FIG. 12, using the same edge synthesis filter e as in the second embodiment.
  • the vertical edge image Sv m is filtered vertically by e, then down-sampled by removing every other pixel in the vertical direction to create a vertical high-frequency image Yv m .
  • the horizontal edge image Sh m is filtered horizontally by e, then down-sampled horizontally to create a horizontal high-frequency image Yh m .
  • the diagonal edge image Sc m is filtered vertically by e, down-sampled in the vertical direction, filtered horizontally by e, then down-sampled in the horizontal direction, creating a diagonal high-frequency image Yc m .
  • the three high-frequency images Yv m , Yh m , and Yc m are all the same size. Each has been down-sampled m times in both the horizontal and vertical directions, reducing the number of pixels by a total factor of 4 m .
  • Inverse wavelet processor 55-m operates on images X' m , Yv m , Yh m , and Yc m as illustrated in FIG. 13, using the same complementary pair of high-pass and low-pass filters g* and h* as in the second embodiment, but not using the completion filter f*.
  • each filtering step is preceded by an up-sampling step and followed by a compensatory doubling of the pixel values. Up-sampling is carried out by inserting zero-valued pixels.
  • Inverse wavelet processor 55-m filters each of the four input images horizontally and vertically, and adds the four resulting images together to obtain the output image X' m-1 .
  • Image X' m is filtered horizontally and vertically by h*;
  • image Yh m is filtered horizontally by g* and vertically by h*;
  • image Yv m is filtered horizontally by h* and vertically by g*;
  • image Yc m is filtered horizontally and vertically by g*
  • the results of the four horizontal filtering operations are added pairwise, so that only two vertical filtering operations have to be performed.
  • inverse wavelet processor 55-m The operation performed by inverse wavelet processor 55-m can also be described by the following equation. As before, if either i/2+k/2 or j/2+1/2 is not an integer, the corresponding values of X' m , Yh m , Yv m , and Yc m are zero. ##EQU18##
  • the third embodiment provides the same advantages as the second and offers improved rendition of high-frequency detail in the reconstructed image, due to diagonal edge detection.
  • the fourth embodiment is similar to the third embodiment, differing only in the operation of the edge detector 52. Block diagrams will be omitted.
  • the edge detector 52 of the fourth embodiment begins by detecting horizontal, vertical, and diagonal edge sharpness by the methods described in the third embodiment, using a threshold value T. It next constructs thickened edges by including contiguous points with edge sharpness values exceeding a lower threshold U. Finally, it trims the edges so that each edge has a uniform thickness (although some edges may be thicker than others). This process will be explained below for an edge in the horizontal edge image Sh.
  • FIG. 14 shows part of a typical horizontal edge image Sh, using stippling to denote pixels with sharpness values exceeding T, and hatching to denote pixels exceeding U but not exceeding T.
  • “exceeding” means having a positive edge sharpness value exceeding T (or U). Pixels with negative sharpness values are processed separately.
  • Edge points in the horizontal edge image Sh tend to be organized into chains with a generally vertical orientation.
  • FIG. 14 shows one such chain, running generally from the top left toward the bottom right.
  • This edge is irregular in shape, however, and if detected by threshold T, the edge is fragmented into two parts. Such irregularities and fragmentation greatly impair the efficiency which which the edge image can be encoded.
  • FIG. 15 shows the thickened version of this edge.
  • the edge detector 52 constructs FIG. 15 from FIG. 14 by the following procedure.
  • the edge detector searches for a pixel in the horizontal edge image Sh in FIG. 14 exceeding threshold T.
  • the first such pixel is found, it is placed in the thickened edge image in FIG. 15 as the first pixel of the first layer of an edge.
  • This first layer is then extended in the left and right directions by including any contiguous pixels having the same sign and exceeding threshold U. The extension continues as far in each direction as such contiguous pixels continue.
  • pixel 56 is found first and pixels 57 and 58 are added to complete the first layer. If there were further pixels exceeding threshold U immediately to the left of pixel 57 or to the right of pixel 58, they would also be added in.
  • the edge detector continues in this manner, extending the thickened edge downward by adding new layers as long as it can find a pixel exceeding threshold T immediately below the bottom layer so far, and enlarging each layer horizontally to include contiguous pixels exceeding threshold T, as well as pixels exceeding threshold U out to the limit extension distance.
  • pixel 60 is not added to the thickened edge because it would extend two pixels beyond the preceding layer, and the extension limit is one pixel.
  • the thickened edge is considered complete.
  • the edge detector searches for another pixel exceeding threshold T in absolute value to start a new thickened edge, and thickens this edge in the same way. This entire procedure is repeated until no more pixels exceeding threshold T can be found.
  • a similar procedure is then followed to process pixels having negative sharpness values exceeding -T and -U.
  • limit extension distance in the example above was one pixel, other limit extension distances such as zero, two, or three pixels can be used.
  • the preferable limit extension distance depends in part on the threshold values T and U.
  • FIG. 16 shows the trimmed version of the thickened edge in FIG. 15.
  • the trimmed edge is constructed by comparing adjacent layers in the thickened edge, and trimming the longer layer to match the length of the shorter layer.
  • pixel 59 is trimmed to reduce the length of its layer to three pixels, matching the length of the layer above. This trimming procedure is carried out from the top of the edge to the bottom, then repeated until all layers in the edge have been trimmed to the same length.
  • the edge detector 52 examines the preceding layer, and trims in such a way as to maintain a constant offset from one layer to the next. In FIG. 16, for example, after trimming layers 61 and 62, the edge detector 52 trims layer 63 so that it is offset one pixel to the right of layer 62, because layer 62 is similarly offset one pixel to the right of layer 61.
  • the trimmed edge in FIG. 16 is continuous, and has a more regular shape than the original edge in FIG. 14. These properties enable the edge in FIG. 16 to be encoded more efficiently than the edge in FIG. 14, even though it contains more pixels.
  • the position of the rest of the edge can be specified just by giving the offset of each layer from the preceding layer.
  • pixels exceeding threshold U but not threshold T can be quantized to zero but then dequantized to a non-zero value in the digital image decoder.
  • edges are thickened and trimmed by a similar procedure, except that the thickening and trimming are done vertically, and the edge detector 52 works from left to right instead of from top to bottom.
  • Edges in the diagonal edge image Sc can be processed either horizontally or vertically. The choice depends on the way in which the diagonal edge image will be encoded. If vertical chain encoding is used, for example, the thickening and trimming should be done horizontally.
  • Edge thickening and trimming can also be added to the edge detection process in the first and second embodiments, and in any of the embodiments that follow.
  • the invented edge thickening and trimming technique is moreover useful for regularizing edges in image encoding and image processing in general, and is not limited to edges detected by the difference-of-differences method.
  • the fifth embodiment is similar to the second embodiment, but is adapted to process moving images comprising successive frames.
  • the digital image encoder of the fifth embodiment has the same two-dimensional smoothing and down-sampling section 41 and edge detector 21 as in the second embodiment, but a different encoding section 65.
  • the encoding section 65 comprises a pair of motion estimators 66 and 67, a motion vector encoder 68, a pair of motion compensators 69 and 70, and a pair of memories 71 and 72. It also has the same edge image encoder 24 and reduced image encoder 25 as in the digital image encoders of the preceding embodiments, and the same edge image decoder 31 and reduced image decoder 32 as in the digital image decoders of the preceding embodiments.
  • Motion estimator 66 receives the pair of edge images Sh and Sv from the edge detector 21. To simplify the drawing, this pair of edge images is denoted by the single letter S. Motion estimator 66 also receives a pair of previous edge images S" (a previous horizontal edge image and a previous vertical edge image) from memory 71. Motion estimator 66 outputs edge motion vectors Ms and a pair of horizontal and vertical edge difference images, denoted for simplicity by the single symbol Ds.
  • Motion estimator 67 receives the reduced image X n from the last smoothing filter 42-n and a previous reduced image X" n from memory 72, and outputs reduced motion vectors Mr and a reduced difference image Dr.
  • the motion vector encoder 68 encodes motion vectors Ms and Mr and outputs motion information Cm.
  • the edge image encoder 24 encodes the edge difference images Ds and outputs edge image information Cs. This information is decoded by the edge image decoder 31 to produce a pair of decoded edge difference images, which are denoted in the drawing by the single symbol D's.
  • Motion compensator 69 receives this pair of images D's, the previous pair of images S", and the motion vectors Ms, and produces a pair of decoded edge images S 1 , which it stores in memory 71.
  • the reduced image encoder 25 encodes the reduced difference image Dr and outputs reduced image information Cr.
  • the reduced image decoder 32 decodes Cr and outputs a decoded reduced difference image D'r.
  • Motion compensator 70 receives this image D'r, the previous reduced image X"n, and the motion vectors Mr, and produces a decoded reduced image X'n which it stores in memory 72.
  • the multiplexer 73 multiplexes the edge image information Cs, motion information Cm, and reduced image information Cr for output as an encoded image C.
  • the digital image decoder of the fifth embodiment has a decoding section 75 comprising a demultiplexer 76, a motion vector decoder 77, and decoders 31 and 32, motion compensators 69 and 70, and memories 71 and 72 identical to those in the digital image encoder in FIG. 17.
  • the demultiplexer 76 receives the encoded image C from the input/output device 28 and demultiplexes it into edge image information Cs, motion information Cm, and reduced image information Cr.
  • the motion vector decoder 77 decodes Cm to obtain decoded edge motion vectors M's and reduced motion vectors M'r.
  • Other symbols in FIG. 18 are as already explained in FIGS. 6 and 17, with S m representing the pair of edge images Sh m and Sv m , and Y m the pair of high-frequency images Yh m and Yv m .
  • motion estimator 67 estimates motion in the reduced image X n by, for example, the well-known block-matching method. This method involves comparing a block of pixels in the image X n with blocks displaced by k pixels horizontally and l pixels vertically from this block in the previous image X" n in memory 72, and selecting the vector (k, l) that gives the closest match. These vectors (k, l) are output as motion vectors Mr, while the differences between the pixel values in X n and the pixels values in X" n at positions offset by these motion vectors (k, l) are output as the reduced difference image Dr.
  • the reduced difference image Dr is the difference between X n and X" n over and above the motion represented by the motion vectors Mr.
  • Motion estimator 66 carries out a similar process separately on the horizontal and vertical edge images to obtain the motion vectors Ms and the pair of edge difference images denoted Ds.
  • the motion vector encoder 68 encodes the motion vectors Ms and Mr by a lossless encoding method such as a variable-length encoding method.
  • the encoders 24 and 25 encode the difference images Ds and Dr by, for example, the same methods as used in the preceding embodiments.
  • the decoders 31 and 32 employ methods reverse to those methods to obtain the decoded difference images D'r and D's.
  • Motion compensator 70 adds the decoded reduced difference image D'r to the previous reduced image X" n according to the motion vectors Mr, and stores the resulting image X' n in memory 72.
  • "According to the motion vectors Mr" means that each pixel value in D'r is added to a corresponding pixel value in the previous reduced image X" n at a location offset by a motion vector (k, l) in Mr.
  • the method by which X' n is obtained from D'r is inverse to the method by which Dr is obtained from X n , and the operations of encoder 25 and decoder 32 are mutually inverse, so image X' n is substantially identical to image X n .
  • motion compensator 69 adds the two decoded edge difference images to the corresponding previous edge images according to the motion vectors Ms, and stores the resulting pair of decoded edge images S 1 in memory 71.
  • the pair of edge images S 1 are substantially identical to the pair of edge images S output by the edge detector 21.
  • Memory 72 has space for storing at least two complete reduced images: the previous image X" n supplied to motion estimator 67 and motion compensator 70, and the new image X' n stored by motion compensator 70. At the end of each frame, the previous image X" n is discarded, leaving the new image X' n to be used as the previous image in the next frame. Memory 71 stores edge images in a similar fashion.
  • the motion estimators 66 and 67 may ignore the previous images S" and X" n in the memories 71 and 72 and supply the edge images S and reduced image X n directly to the encoders 24 and 25, instead of supplying difference images.
  • the motion information Cm in this case consists of a message indicating that the current frame does not require motion compensation.
  • the motion compensators 69 and 70 then simply output the images received from the decoders 31 and 32.
  • the motion compensators 69 and 70 and memories 71 and 72 in the digital image decoder operate in the same way as the motion compensators 69 and 70 and memories 71 and 72 in the digital image encoder, using the decoded motion vectors M's and M'r supplied by the motion vector decoder 77. Since a lossless encoding scheme is employed for the motion vectors, M's and M'r are exactly identical to Ms and Mr, and the images denoted S 1 and X' n in FIG. 18 are exactly identical to the images denoted S 1 and X' n in FIG. 17. Further operation of the digital image decoder is as already described in the second embodiment.
  • One advantage of the fifth embodiment is that most of the block matching is performed in the edge images, which have been thresholded and tend to consist mostly of zero-valued pixels.
  • the large number of zero-valued pixels greatly speeds up the block-matching process. Further gains in speed are possible by using motion vectors calculated for one of the two edge images to help find motion vectors in the other edge image.
  • Another advantage is that the motion estimators and compensators in the digital image encoder and digital image decoder use identical previous images S" and X" n . This prevents the accumulation of quantization error and other image errors resulting from compressive coding in the encoders 24 and 25, and ensures consistency between motion estimation in the digital image encoder and motion compensation in the digital image decoder.
  • the fifth embodiment can be employed in video teleconferencing and a wide variety of other multimedia applications.
  • the sixth embodiment is similar to the fifth embodiment, with further provisions for speeding up the computation of motion vectors.
  • the encoding section 78 of the digital image encoder in the sixth embodiment is similar to the encoding section 65 in the fifth embodiment, but has a bidirectional interconnection 79 between motion estimators 66 and 67.
  • This enables motion estimator 67 to receive the edge motion vectors Ms found by motion estimator 66, scale them down to the size of the low-frequency reduced image X n , and restrict the search for motion vectors in image X n to vectors close to the scaled-down edge motion vectors.
  • the search can be eliminated entirely: the reduced motion vectors Mr can be computed directly from the edge motion vectors Ms.
  • motion estimator 66 can receive the motion vectors Mr found by block matching in the low-frequency reduced image X n , scale them up to the size of the edge images Sh and Sv, and restrict the search for motion vectors in Sh and Sv to vectors close to the scaled-up vectors. This scheme is particularly effective in detecting large, rapid motion.
  • the sixth embodiment has the same structure as the fifth embodiment and operates in the same way, so further description will be omitted.
  • the sixth embodiment enables frames to be encoded at a faster rate than in the fifth embodiment.
  • the seventh embodiment combines features of the third and fifth embodiments by detecting edges horizontally, vertically, and diagonally, and employing motion compensation.
  • the digital image encoder of the seventh embodiment has the same two-dimensional smoothing and down-sampling section 41, one-dimensional smoothing and down-sampling section 51, and edge detector 52 as in the third embodiment (FIG. 8), and the same encoding section 65 as in the fifth embodiment (FIG. 17).
  • the symbol S in FIG. 20 denotes the set of three edge images Sv, Sh, and Sc in FIG. 8.
  • Each of the symbols S 1 , S", Ds, and D's likewise denotes a set of three edge images.
  • the digital image decoder of the seventh embodiment has the same decoding section 75 as the fifth embodiment (FIG. 17), and the same two-dimensional smoothing and down-sampling filters 53-1 to 53-(n-1), edge synthesizers 54-1 to 54-n, and inverse wavelet transform processor 55 as in the third embodiment (FIG. 11).
  • the seventh embodiment operates as explained in the third and fifth embodiments, and provides similar effects, so further description will be omitted.
  • the eighth embodiment combines features of the sixth and seventh embodiments.
  • the digital image encoder of the eighth embodiment has the same configuration as in the seventh embodiment (FIG. 20), except for the addition of a bidirectional interconnection 79 between motion estimators 66 and 67 as in the sixth embodiment (FIG. 19).
  • the eighth embodiment operates as explained in the sixth and seventh embodiments, so further description will be omitted.
  • the ninth embodiment performs motion estimation before edge detection, smoothing, and down sampling.
  • the encoder performs substantially all the processes carried out in the decoder, including edge synthesis and the inverse wavelet transform.
  • the original image frames input at the input terminal 10 are supplied to a motion estimator 80, which performs motion estimation as described above, outputting a difference image Do and motion vectors M.
  • the motion vectors M are encoded by a motion vector encoder 68 as in the fifth to eighth embodiments.
  • the difference image D 0 is supplied to an edge detector 21 and two-dimensional smoothing and down-sampling section 41 in place of the original image X 0 , and is processed in the same way that the original image X 0 was processed in the second embodiment.
  • the resulting pair of edge images will again be denoted by the symbol S, while the reduced image resulting from smoothing and down-sampling of D 0 will be denoted D n .
  • These images S and D n are encoded by respective encoders 24 and 25 to produce encoded image information Cs and Cr, which is combined with the motion information Cm by a multiplexer 73 as in the preceding embodiments.
  • the inverse wavelet transform processor 50 outputs a reconstructed difference image D' 0 to a motion compensator 81.
  • the motion compensator 81 uses this image D' 0 , the motion vectors M output by the motion estimator 80, and a previous image X" 0 stored in a memory 82, the motion compensator 81 produces a new reconstructed image X' 0 , which is stored in the memory 82.
  • the previous image X" 0 was also supplied to the motion estimator 80 for use in motion estimation.
  • the reconstructed image X' 0 will become the previous image in the next frame.
  • the motion estimator 80, motion compensator 81, and memory 82 in FIG. 23 operate like the motion estimator 67, motion compensator 70, and memory 72 in the fifth or seventh embodiment (FIG. 17 or 20), although they operate on the original image X 0 instead of the reduced image X n .
  • the encoding section 83 in the ninth embodiment comprises the motion vector encoder 68, edge image encoder 24, reduced image encoder 25, and multiplexer 73.
  • the digital image decoder of the ninth embodiment has a decoding section 84 with a demultiplexer 76 and motion vector decoder 77 similar to those in the fifth to eighth embodiments.
  • the other elements in FIG. 24 are similar to the elements with the same reference numerals in FIG. 23.
  • the motion compensator 81 receives a fully reconstructed difference image D' 0 from the inverse wavelet transform processor 50, and outputs a fully reconstructed image X' 0 to the input/output device 37.
  • the digital image encoder in FIG. 23 and digital image decoder in FIG. 24 operate with identical previous images X" 0 , thereby maintaining consistency between motion estimation in the encoder and decoder. Performing motion estimation before edge detection can lead to higher compression ratios than in the fifth to eighth embodiments, because only one set of motion vectors has to be encoded.
  • the tenth embodiment combines features of the third and ninth embodiments.
  • Three edge images are detected, as in the third embodiment.
  • Edge detection, smoothing, and down-sampling are preceded by motion estimation, and followed by all the processes performed in the digital image decoder, as in the ninth embodiment.
  • the digital image encoder of the tenth embodiment has the motion estimator 80, motion compensator 81, memory 82, and encoding section 83 of the ninth embodiment (FIG. 23), and other elements as in the third embodiment (FIGS. 9 and 11).
  • the digital image decoder has the decoding section 84, motion compensator 81, and memory 82 of the ninth embodiment (FIG. 24), and other elements as in the third embodiment (FIG. 11).
  • the tenth embodiment operates as described in the third and ninth embodiments, and provides similar advantages. A detailed description will be omitted.
  • the eleventh embodiment is a digital image encoder/decoder (codec) that combines the functions of the digital image encoder and decoder of the ninth embodiment.
  • codec digital image encoder/decoder
  • the eleventh embodiment thus provides one device that can operate in either an encoding mode or a decoding mode.
  • the eleventh embodiment adds the demultiplexer 76 and motion vector decoder 77 of the digital image decoder of the ninth embodiment (FIG. 24), a second input terminal 90, and three switches 91, 92, and 93.
  • switches 91, 92, and 93 are set to the position marked "a" in the drawing.
  • Successive frames X 0 are input at input terminal 10 and processed just as in the digital image encoder of the ninth embodiment, the encoded results being output from the demultiplexer 73 to input/output device 28.
  • switches 91, 92, and 93 are set to the position marked "b."
  • the encoded moving image C is input from the second input terminal 90 to the demultiplexer 76 and processed just as in the digital image decoder of the ninth embodiment, the decoded results being output from the motion compensator 81 to input/output device 37.
  • the eleventh embodiment has the same advantages as the ninth embodiment, with the further advantage of allowing the encoding and decoding functions to share the same motion compensator 81, decoders 31 and 32, filters 48-m, edge synthesizers 49-m, and inverse wavelet transform processor 50, thereby avoiding needless duplication.
  • This embodiment is particularly suited for videophones and other bidirectional video systems.
  • the twelfth embodiment is a digital image encoder/decoder that combines the features of the tenth and eleventh embodiments.
  • the twelfth embodiment adds to the digital image encoder of the tenth embodiment (FIG. 25) the same demultiplexer 76, motion vector decoder 77, second input terminal 90, and switches 91, 92, and 93 as in the eleventh embodiment.
  • the twelfth embodiment operates like the eleventh, but has the further advantage of improved rendition of high-frequency image detail due to diagonal edge detection.
  • the invented encoding and decoding method and devices provide high compression rates with modest amounts of computation. Although there is some loss of detail, reconstructed images are not distorted, and because high-frequency information is synthesized in a single step instead of by iteration, the reconstruction process is reliably completed within a short and definite time.
  • the invention can be usefully applied to both still and moving images for a wide variety of purposes, in a wide variety of systems.
  • the edge detector 21 can output a single edge image in which each edge point has both a sharpness value S(i, j) and direction angle ⁇ (i, j), computed from Sh(i, j) and Sv(i, j) by, for example, the following formulas:
  • sgn denotes the algebraic sign (plus or minus one), and arctan the inverse tangent function.
  • the encoding section then has only one set of edge positions to encode.
  • the decoding section in the digital image decoder can recover Sh(i, j) and Sv(i, j) from S(i, j) and ⁇ (i, j).
  • This method of encoding only a single edge image is particularly effective if chain encoding is employed, because the direction angle tends to change only slowly along a chain of edge points. Moreover, the angle tends to be perpendicular to the direction of the chain, so it is possible to have the edge detector calculate only the above edge sharpness S(i, j), and let the decoder infer the direction angle ⁇ (i, j) of an edge point (i, j) from the direction of the chain in the vicinity of that point.
  • the compression ratio can be significantly improved in this way, although with some loss of reconstruction accuracy.
  • This encoding method can also be applied to the horizontal and vertical edge images in the other embodiments, if the original image is not down-sampled before horizontal and vertical edge detection. The necessary down-sampling is then performed instead in the decoder, as in the second embodiment.
  • this encoding method has the further advantage of reducing the number of edge images for which motion vectors have to be computed.
  • the two-dimensional smoothing performed in a single step, instead of by a cascaded series of filters.
  • the filters h 1 , h 2 , . . . , h n in the first embodiment, for example, can be mutually convolved to yield a single filter that gives the same result as applying the cascaded series of filters.
  • down-sampling can be incorporated into this single filter simply by calculating the values of only 1/4 n of the output pixels.
  • the smoothing filters applied in the horizontal and vertical directions need not be identical.
  • a smoothing filter with an even number of taps can be applied in one of these directions, and a smoothing filter with an odd number of taps in the other direction.
  • the edge synthesis filters (e), completion filters (f), and high-pass filters (g) employed for decoding then also require separate versions for the horizontal and vertical directions.
  • Edge detection is not limited to the difference-of-differences method. This method is preferred, but other methods can be employed.
  • the filters employed were described as being symmetric around zero or 1/2, but this is only a matter of notation.
  • the center of symmetry can be disposed at any convenient location.
  • the standard edge from which the edge synthesis filters are generated need not be an edge of the type shown in FIG. 3, in which pixel values change at first one constant rate, then another constant rate. Other types can be used, provided the standard edge has only one point which would be detected as an edge point by the edge detector.

Abstract

A digitized image is encoded by detecting edges in the image, encoding the position and sharpness of the detected edges, smoothing and reducing the size of the digitized image, and encoding the resulting reduced image. A digitized image encoded in this way is reconstructed by generating a horizontal edge image and a vertical edge image, or a horizontal edge image, a vertical edge image, and a diagonal edge image, from the encoded edge position and sharpness information, smoothing each of these edge images by a series of filters to obtain a series of edge images, filtering each edge image in each series with an edge synthesis filter to synthesize a high-frequency image, decoding the reduced image, and performing an inverse wavelet transform on the decoded reduced image and the high-frequency images.

Description

CROSS REFERENCE TO RELATED APPLICATIONS
This is a division under 35 U.S.C. 120 of parent application, Ser. No. 08/548,943, filed Oct. 26, 1995 now U.S. Pat. No. 5,761,341, the disclosure of which is incorporated herein by reference.
BACKGROUND OF THE INVENTION
This invention relates to a compressive image encoding and decoding method using edge synthesis and the inverse wavelet transform, and to digital image encoding and decoding devices employing this method.
Compression is essential for efficient storage and transmission of digitized images. Compression methods have been described by the Joint Photographic Experts Group (JPEG) for still images, and the Motion Picture Experts Group (MPEG) for moving images. The JPEG method involves a discrete cosine transform (DCT), followed by quantization and variable-length encoding. The MPEG method involves detecting motion vectors. Both methods require substantial computation, the detection of motion vectors being particularly demanding.
Recently there has been much interest in the wavelet transform as a means of obtaining high compression ratios with relatively modest amounts of computation. This transform employs a family of wavelets related by dilation and translation; that is, the family consists of occurrences of the same basic wavelet at different locations and on different scales. If the scales form a progressively doubling sequence, and if the basic wavelet is zero everywhere except in a limited domain, wavelet transforms and inverse wavelet transforms can be carried out with efficient computational algorithms.
A wavelet transform can be described as a filtering process executed at each wavelet scale. A digitized image, for example, is transformed by filtering with the basic wavelet, then with the basic wavelet dilated by a factor of two, then with the basic wavelet dilated by a factor of four, and so on.
One prior-art wavelet encoding scheme employs a complementary pair of wavelets to divide an image into a high-frequency component and a low-frequency component. These components contain information about variations on scales respectively less than and greater than a certain cut-off scale. This process is iterated on the low-frequency component with a doubling of the wavelet scale, obtaining new low-frequency and high-frequency components, then iterated again on the new low-frequency component, and so on. After a certain number of iterations, the components are encoded by an encoding scheme that works from low-toward high-frequency information. This scheme enables accurate image reconstruction, but retains too much high-frequency information to achieve high compression ratios.
Another prior-art wavelet encoding scheme employs a basic wavelet that is the first derivative of a smoothing filter (that is, the first derivative of a low-pass filtering function). This type of wavelet acts as a high-pass filter. High-frequency information is obtained by detecting local peaks (local maxima of absolute values) in the result of the wavelet transform, which correspond to edges in the original image. The size and location of the peak values at a selected scale are encoded, along with a low-frequency image obtained by smoothing at the largest scale of the wavelet transform. Fairly high compression ratios can be obtained in this way.
To reconstruct the original image from the encoded data, this prior-art method employs an algorithm derived from a mathematical procedure involving iterated projections in Hilbert space. Under ideal conditions, the projections converge toward a unique set of data that (i) have the required local peak values and (ii) are within the range of the wavelet transform operator. An inverse wavelet transform is then carried out on the converged data to obtain the original image.
It has yet to be shown, however, that the projections always converge, or that data satisfying conditions (i) and (ii) are unique. In practice, there is difficulty in knowing when to stop iterating. For some images, it seems that data satisfying (i) and (ii) are not unique, and instead of converging, the iteration wanders endlessly through Hilbert space, first approaching the desired image transform, then moving away again.
A further problem of edge-based encoding schemes in general is that encoding efficiency tends to be impaired by edge fragmentation and irregularity.
SUMMARY OF THE INVENTION
It is accordingly an object of the present invention to encode digitized images efficiently, with a high compression ratio but without requiring extensive computation.
Another object of the invention is to encode moving images efficiently.
Still another object is to reduce irregularity of edges.
Yet another object is to avoid fragmentation of edges.
Still another object is to reconstruct digitized images from encoded information efficiently and reliably.
Yet another object is to reconstruct digitized moving images efficiently and reliably.
The invented method of encoding a digitized image comprises the steps of:
detecting edges in the digitized image;
encoding the position and sharpness of the detected edges to generate edge image information;
filtering and down-sampling the digitized image to
generate a reduced image; and
encoding the reduced image to generate reduced image information.
Detecting edges in the digitized image may comprise the further steps of thickening the detected edges, then trimming each edge to a uniform thickness.
The invented method of reconstructing a digitized image encoded by the invented encoding method comprises the steps of:
generating a set of decoded edge images from the above edge image information;
filtering these decoded edge images to generate a series of sets of decoded edge images that are increasingly smooth, or increasingly small in size;
synthesizing a series of sets of high-frequency images by filtering the decoded edge images with respective edge synthesis filters;
decoding the reduced image information to obtain a decoded reduced image; and
performing an inverse wavelet transform on the decoded reduced image and the above sets of high-frequency images.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of the digital image encoder in a first embodiment of the invention.
FIG. 2 is a block diagram of the digital image decoder in the first embodiment.
FIG. 3 is a graph illustrating pixel values at an edge.
FIG. 4 is a block diagram of the digital image encoder in a second embodiment of the invention.
FIG. 5 is flow diagram illustrating the smoothing and down-sampling process in the encoder of the second embodiment.
FIG. 6 is a block diagram of the digital image decoder in the second embodiment.
FIG. 7 is a flow diagram illustrating the synthesis of high-frequency images in the second embodiment.
FIG. 8 is a flow diagram illustrating the inverse wavelet transform in the second embodiment.
FIG. 9 is a block diagram of the digital image encoder in a third embodiment of the invention.
FIG. 10 is a flow diagram illustrating one-dimensional filtering and down-sampling in the third embodiment.
FIG. 11 is a block diagram of the digital image decoder in the third embodiment.
FIG. 12 is a flow diagram illustrating the synthesis of high-frequency images in the third embodiment.
FIG. 13 is a flow diagram illustrating the inverse wavelet transform in the third embodiment.
FIG. 14 is a diagram of an edge, illustrating irregular shape and fragmentation.
FIG. 15 illustrates thickening of the edge in FIG. 14 according to a fourth embodiment of the invention.
FIG. 16 illustrates trimming of the edge in FIG. 15 according to the fourth embodiment.
FIG. 17 is a block diagram of the digital image encoder in a fifth embodiment, for encoding moving images.
FIG. 18 is a block diagram of the digital image decoder in the fifth embodiment.
FIG. 19 is a block diagram of the digital image encoder in a sixth embodiment of the invention.
FIG. 20 is a block diagram of the digital image encoder in a seventh embodiment.
FIG. 21 is a block diagram of the digital image decoder in the seventh embodiment.
FIG. 22 is a block diagram of the digital image encoder in an eighth embodiment.
FIG. 23 is a block diagram of the digital image encoder in a ninth embodiment.
FIG. 24 is a block diagram of the digital image decoder in the ninth embodiment.
FIG. 25 is a block diagram of the digital image encoder in a tenth embodiment.
FIG. 26 is a block diagram of the digital image decoder in the tenth embodiment.
FIG. 27 is a block diagram of a digital image encoder/decoder in an eleventh embodiment of the invention.
FIG. 28 is a block diagram of a digital image encoder/decoder in a twelfth embodiment.
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the invention will be described with reference to the attached illustrative drawings. The following terminology will be used.
A (two-dimensional) digitized image is an array of pixels having values x(i, j), where the horizontal coordinate i and vertical coordinate j range over sets of integers. The range of these coordinates is the size of the image, i.e., the number of pixels in the horizontal and vertical directions. The pixel values represent, for example, intensity gradations.
A one-dimensional image is a horizontal or vertical line in a two-dimensional digitized image, i.e., a set of pixels with values x(i) indexed by a single coordinate.
A standard edge is a one-dimensional image having just one sharp variation point. An example would be an image with pixel values that change first at one constant rate, then at a different constant rate, such as the following:
. . . 2.0 1.5 1.0 0.5 0 0.5 1.0 1.5 2.0 . . .
In this standard edge the pixel values first decrease at a rate of -0.5 per pixel, then increase at a rate of +0.5 per pixel. A standard edge serves as a basic model for all the edges occurring in a digitized image.
Down-sampling a digitized image means reducing its size by a process generally analogous to photographic reduction: by selecting every n-th pixel, for example, where n is a suitable integer. Up-sampling means increasing the size of an image by interpolating new pixels between the existing pixels.
A (one-dimensional) filter is a set of coefficients f(k), where k ranges over the integers. The non-zero coefficients f(k) are the taps of the filter.
The Fourier transform F of a filter f is defined as follows, where summation is over all taps, e is the natural logarithm base, and j is the square root of minus one. ##EQU1##
The conjugate f* of a filter f is obtained by reversing the order of coefficients:
f*(k)=f(-k)
The Fourier transform of f* is therefore the complex conjugate of the Fourier transform of f.
A filter has even symmetry around i0 if it has equal values on both sides of i0 ; that is, for all values of k:
f(i.sub.0 -k)=f(i.sub.0 +k)
In particular, a filter has even symmetry around zero if it is equal to its own conjugate.
A filter has odd symmetry around i0 if it has opposite values on both sides of i0, that is:
f(i.sub.0 -k)=-f(i.sub.0 +k)
A filter satisfies the exact reconstruction condition if its Fourier transform F satisfies the following condition for all values of ω:
|F(ω)|.sup.2 +|F(ω+π)|.sup.2 =1
Two filters with Fourier transforms G and H satisfy the exact reconstruction condition if:
|G(ω)|.sup.2 +|H(ω)|.sup.2 =1
Filters satisfying these exact reconstruction conditions, also referred to as perfect reconstruction conditions, are well known in the wavelet transform art. A pair of filters satisfying the second exact reconstruction condition above is referred to as a complementary pair.
To filter a one-dimensional image x by a filter f means to obtain a new one-dimensional image y with the following pixel intensities: ##EQU2##
This operation is equivalent to the conventional convolution operation using the conjugate filter f*. ##EQU3## The difference between convolution and filtering, as the term is used herein, is purely one of notation. The meaning of the following description and claims would be unchanged if the terms "convolution" and "convolving" were used throughout, instead of "filtering."
To filter a digitized image x horizontally by a filter f means to obtain a new image y as follows: ##EQU4##
Similarly, filtering an image x vertically by f means: ##EQU5##
Filtering can be done two-dimensionally by, for example, filtering first in the horizontal direction, then in the vertical direction, or vice versa.
A high-pass filter is a filter that retains small-scale variations and rejects large-scale variations. A low-pass filter rejects small-scale variations and retains large-scale variations. Smoothing filter is a synonym for low-pass filter.
The cut-off frequency of a high-pass or low-pass filter indicates the scale that divides the retained variations from the rejected variations. A lower cut-off frequency corresponds to a larger scale of variations. Frequency actually refers to the variable ω in the Fourier transform F(ω) of the filter.
Generally speaking, the Fourier transform of a high-pass filter satisfies F(0)=0, while for a low-pass filter, |F(0)|>0. If |F(0)|=1, a low-pass filter can retain large-scale variations without changing their size. Incidentally, F(0) is the sum of the filter coefficients f(k).
First embodiment
Referring to FIG. 1, the digital image encoder in the first embodiment comprises an input terminal 10, a two-dimensional smoothing and down-sampling section 20, and an edge detector 21. The two-dimensional smoothing and down-sampling section 20 comprises a cascaded series of two-dimensional smoothing filters 22-m (m=1 to n, where n is an integer greater than one), and a down sampler 23. The digital image encoder also comprises an edge image encoder 24, a reduced image encoder 25, and a multiplexer 26, which together constitute an encoding section 27.
A digitized image X0 is input from the input terminal 10 to the edge detector 21 and first smoothing filter 22-1. Smoothing filter 22-1 outputs a filtered image X1 to the next smoothing filter 22-2, and this process continues, each smoothing filter 22-m filtering the image Xm-1 received from the preceding smoothing filter and furnishing a new filtered image Xm to the next smoothing filter, until the last smoothing filter 22-n is reached. The last smoothing filter 22-n outputs a low-frequency image Xn to the down sampler 23, which outputs a reduced image R to the reduced image encoder 25. The reduced image encoder 25 encodes the reduced image R to obtain reduced image information Cr.
The edge detector 21 detects edge points in the input image X0 and outputs horizontal and vertical edge images Sh and Sv to the edge image encoder 24 in the encoding section 22. The edge image encoder 24 encodes these edge images to obtain edge image information Cs. The multiplexer 26 combines the reduced image information Cr and edge image information Cs into an encoded image C, which is output to a first input/output device 28 such as a communication channel, semiconductor memory, or magnetic disk storage device.
Referring to FIG. 2, the digital image decoder has a decoding section 29 comprising a demultiplexer 30, an edge image decoder 31, and a reduced image decoder 32. The digital image decoder also has a cascaded series of two-dimensional smoothing filters 33-m(m=1 to n-1), a series of edge synthesizers 34-m (m=1 to n), an up sampler 35, and an inverse wavelet transform processor 36. The inverse wavelet transform processor 36 comprises a cascaded series of inverse wavelet processors 36-m (m=1 to n).
The demultiplexer 30 receives the encoded image C from the first input/output device 28 and separates it into edge image information Cs, which it supplies to the edge image decoder 31, and reduced image information Cr, which it supplies to the reduced image decoder 32. The edge image decoder 31 decodes Cs to obtain a pair of horizontal and vertical edge images Sh1 and Sv1. These decoded edge images are filtered in turn by the smoothing filters 33-m (m=1 to n-1), producing a series of pairs of edge images Shm and Svm (m=1 to n) in which the unfiltered edge images Sh1 and Sv1 are the first pair.
Each edge synthesizer 34-m receives the corresponding pair of edge images Shm and Svm, and synthesizes a pair of horizontal and vertical high-frequency images Yhm and Yvm (m=1 to n). The reduced image decoder 32 decodes Cr to obtain a decoded reduced image R', which the up sampler 35 enlarges to obtain a decoded low-frequency image X'n.
In the inverse wavelet transform processor 36, each inverse wavelet processor 36-m (m=1 to n) receives the corresponding high-frequency images Yhm and Yvm and a partially reconstructed image X'm from the preceding inverse wavelet processor 36-(m+1) in the cascade, and outputs a partially reconstructed image X'm-1 to the next inverse wavelet processor 36-(m-1). The first inverse wavelet processor 36-n in the cascade receives the decoded low-frequency image X'n from the up sampler 35. The last inverse wavelet processor 36-1 outputs the fully reconstructed image X'0 to a second input/output device 37.
Each of the elements in the FIGS. 1 and 2 comprises a memory for storing data and well-known circuits for performing arithmetic and logic operations. Descriptions of the circuit configurations will be omitted to avoid obscuring the invention with irrelevant detail. The entire digital image encoder can be fabricated as a single semiconductor integrated circuit, or can be built from a small number of such circuits; likewise the digital image decoder. The invention can also be practiced by providing a general-purpose processor such as a microprocessor or digital signal processor with memory for storing data and programs for executing the functions of the individual elements in FIG. 1 or FIG. 2.
Next the operation will be described, starting with the operation of the encoder in FIG. 1.
The input image X0 is a digitized image with pixel values X0 (i, j). The edge detector 21 detects edges by detecting variations in the rate of change of pixel values from pixel to pixel. More specifically, the edge detector 21 takes differences between adjacent pixel values, then takes differences between these differences, and finally performs a thresholding operation to reduce small differences to zero. This process is carried out separately in the horizontal and vertical directions.
In the horizontal direction, for each pixel, the edge detector 21 computes the following pixel difference Pdh(i, j):
Pdh(i, j)=X.sub.0 (i, j)-X.sub.0 (i-1, j)
Next it computes differences Sh(i, j) of these pixel differences Pdh(i, j) as follows: ##EQU6##
Equivalent edge detection could be performed by filtering the image X0 with a second-derivative filter having coefficients (1, -2, 1), but taking differences of differences is preferable because it requires less computation.
The detected value Sh(i, j) is the edge sharpness at pixel (i, j), also referred to as the edge size. Edge sharpness values with absolute values equal to or less than a certain threshold value T are quantized to zero. Values exceeding T are also quantized, to reduce the number of bits in the edge-image data; then the quantized sharpness values Sh(i, j) at all pixels are output as the horizontal edge image Sh. Non-zero pixels in the edge image, i.e. pixels at which |Sh(i, j)|>T, are referred to as edge points.
The meaning of the edge sharpness computation can be seen in FIG. 3. The horizontal axis represents, for example, the horizontal direction in the image, and the vertical axis represents the pixel value. The difference between pixel values 38 and 39 is -A, the difference between pixel values 40 and 38 is B, and the difference between these differences is B-(-A)=A+B. If A+B>T, then pixel 38 is an edge point with sharpness A+B. The other pixels in FIG. 3 have zero sharpness. FIG. 3 thus illustrates a standard edge, with just one sharp variation.
The vertical edge image is generated in the same way, by taking differences in the vertical direction, then taking differences of these differences. The edge sharpness Sv in the vertical direction is:
Sv(i, j)=X.sub.0 (i, j+1)-2X.sub.0 (i, j)+X.sub.0 (i, j-1)
The vertical edge image is quantized as was the horizontal edge image, points for which |Sv(i, j)|<T being set to zero. Quantization can be performed by dividing the edge sharpness values by a power of two, for example, or by using a quantization table. Incidentally, the same notation Sh and Sv will be employed herein to denote edge sharpness values both before and after quantization.
Smoothing filter 22-1 employs a low-pass filter h1 with even symmetry and a certain number of taps. In the following description it will be assumed that this filter either has 2N+1 taps and is symmetric around zero, or has 2N taps and is symmetric around 1/2, where N is an appropriate positive integer. The even symmetry condition in the first case is:
h.sub.1 (-k)=h.sub.1 (k) (k=1, 2, . . . , N)
The even symmetry condition in the second case is:
h.sub.1 (1-k) h.sub.1 (k) (k=1, 2, . . . , N)
The low-pass filter h1 satisfies the exact reconstruction condition, and the sum of its tap coefficients is plus or minus one. That is, the Fourier transform H1 of h1 satisfies the following conditions:
|H.sub.1 (ω)|.sup.2 + |H.sub.1 (ω + π)|.sup.2 = 1
|H.sub.1 (0)| = 1
The smoothing filter 22-1 performs a two-dimensional low-pass filtering operation by filtering the input image X0 with h1 horizontally and vertically, obtaining: ##EQU7##
The other smoothing filters 22-m (m>1) also employ low-pass filters hm with even symmetry, and with tap coefficients summing to plus or minus one. The cut-off frequency of each smoothing filter 22-m (m>1) is one-half the cut-off frequency of the preceding filter 22-(m-1) in the cascade. Filters with these cut-off frequencies can be obtained by doubling the tap spacing from one filter to the next, and inserting zeros to fill the vacated spaces. If the number of taps is odd, for example, then hm can be obtained from hm-1 as follows:
h.sub.m (2k) = h.sub.m-1 (k)
h.sub.m (2k + 1) = 0 (k = 0, ±1, ±2, . . . )
In smoothing filter 22-m, image Xm-1 is filtered horizontally and vertically by filter hm to obtain image Xm. As the image passes through the cascade, variations on progressively larger scales are smoothed out, and resolution of fine detail decreases correspondingly.
Incidentally, if the low-pass filters hm (m>1) are obtained by doubling the tap spacing as described above, they will pass high-frequency variations in certain regions above their cut-off frequencies. This is allowable because those high-frequency variations will already have been removed by previous filters in the cascade.
The down sampler 23 down-samples the low-frequency image Xn output by the smoothing filter 22-n, thereby reducing the amount of image data. One simple method of down-sampling is to divide the image into blocks of M×M pixels each, and replace each block by a representative pixel value such as the mean or median value in the block. This reduces the size of the image by a factor of M in both the horizontal and vertical directions, and reduces the amount of image data by a factor of M2. (M is an appropriate integer, such as M=2n.) The reduced image R output by the down sampler 23 resembles a photographic reduction of the original image X0.
The edge image encoder 24 encodes the quantized horizontal and vertical edge images Sh and Sv by using, for example, run-length encoding to encode the positions of edge points, and differential encoding to encode the sharpness values at these points. Alternatively, chain encoding can be used to encode the edge positions, taking advantage of the fact that edge points tend to be organized into chains of mutually contiguous points. Chain encoding encodes only the displacement from one position to the next in such a chain.
The reduced image encoder 25 encodes the reduced image R by, for example, performing a discrete cosine transform, quantizing the resulting DCT coefficients, and encoding them in zig-zag order by a variable-length encoding method. Alternatively, the reduced image R can be encoded by a predictive encoding method similar to the differential pulse-code modulation method (DPCM) commonly employed for audio signals, or by any other suitable method. If the above value of M is sufficiently large, high coding efficiency is not critically important, because the amount of data to be encoded has already been greatly reduced by down-sampling.
The encoded image C, consisting of the edge image information Cs and low-frequency information Cr as multiplexed by the multiplexer 26, is output to the first input/output device 28 for transmission or storage. The amount of encoded data is generally reduced to a small fraction of the amount in the original image X0, so the encoded image C can be efficiently transmitted or stored.
Next the decoding operations will be described. These operations employ further filters em, fm, and gm.
Filter g1 is a high-pass filter related to the smoothing filter h1 as follows:
g.sub.1 (k)=(-1).sup.k h.sub.1 (-k)
This relation implies that g1 has even symmetry if h1 was symmetric around zero, or odd symmetry if h1 was symmetric around 1/2. It further implies that the Fourier transform G1 of g1 is the complex conjugate of H1 (ω+π), so in particular, h1 and g1 satisfy the exact reconstruction condition and form a complementary pair. Furthermore, G1 (ω+π) is the complex conjugate of H(ω), so g1 also satisfies an exact reconstruction condition by itself:
|H.sub.1 (ω)|.sup.2 + |G.sub.1 (ω)|.sup.2 = 1
|G.sub.1 (ω)|.sup.2 + |G.sub.1 (ω + π)|.sup.2 = 1
Since |H1 (0)|=1, it follows that |G1 (0)|=0, so the sum of the tap coefficients g1 (k) is zero.
The other filters gm (m=2 to n) are also high-pass filters, and are related to the corresponding hm by a similar equation:
|H.sub.m (ω)|.sup.2 +|G.sub.m (ω)|.sup.2 =1
These filters gm can be obtained from g1 by successively doubling the tap spacing, in the same way that the filters hm were obtained from h1.
Filter fm is a completion filter, defined in terms of its Fourier transform Fm as follows. ##EQU8##
Filter fm is a low-pass filter, referred to as a completion filter, obtained by executing an inverse Fourier transform on Fm. Once f1 has been obtained, the other fm (m=2 to n) can be derived from it by successively doubling the tap spacing. The above definition of Fm, incidentally, is well known in the wavelet transform art and comes from the following equation:
|H(ω)|.sup.4 +2·F(ω)·|Gω)|hu 2=1
Filter em is an edge synthesis filter derived by filtering a standard edge of unit sharpness by the corresponding high-pass filter gm. When gm has even symmetry, all standard edges with unit sharpness lead to the same em, which has the following tap values em (t). L is the half-length of gm, i.e. the distance from the central tap to the outermost tap. If gm is derived as described above, then L=2(m-1)N). ##EQU9##
When gm has odd symmetry, the situation is a little more complicated and will be described later.
Referring again to FIG. 2, after the demultiplexer 30 separates the input encoded image C into edge image information Cs and reduced image information Cr, decoders 31 and 32 carry out processes that are reverse to the processes performed by encoders 24 and 25 in FIG. 1. For example, if the reduced image encoder 25 in FIG. 1 performed a discrete cosine transform followed by quantization and variable-length encoding, the reduced image decoder 32 in FIG. 2 performs variable-length decoding, dequantization, then an inverse discrete cosine transform to obtain the decoded reduced image R'.
If the edge images were encoded by run-length or chain encoding of the positions of the edge points and differential encoding of their sharpness, the edge image decoder 31 performs run-length or chain decoding to recover the edge positions and an accumulation process to obtain the sharpness values. The edge image decoder 31 also dequantizes the sharpness values.
The up sampler 35 obtains the decoded low-frequency image X'n by restoring the decoded reduced image R' to the original image size (the size of the edge images and high-frequency images). If, for example, each pixel value in the reduced image R represented an M×M block in the low-frequency image Xn, the up sampler 35 can simply copy this value to all pixels in the corresponding M×M block of the decoded low-frequency image X'1. More sophisticated up-sampling methods involving interpolation can also be employed, but smooth interpolation is not required, because the image will be smoothed during the inverse wavelet transform.
The cascaded smoothing filters 33-m (m=1 to n-1) carry out successive two-dimensional filtering operations on the pairs of edge images Shm and Svm in the same way that the cascaded smoothing filters 22-m carried out successive filtering operations in the encoder. Specifically, Shm is filtered by hm horizontally and vertically to obtain Shm+1, and Svm is filtered by hm horizontally and vertically to obtain Svm+1.
Each edge synthesizer 34-m filters the corresponding horizontal edge image Shm by the edge synthesis filter em in the horizontal direction, and the vertical edge image Svm by em in the vertical direction. When the smoothing filters hm have an odd number of taps and the high-pass filters gm accordingly have even symmetry, the resulting horizontal and vertical high-frequency images Yhm and Yvm can be described by the following equations: ##EQU10##
Conceptually, for each edge point, the edge synthesizer 34 synthesizes the result of filtering a standard edge of the corresponding sharpness by the high-frequency filter gm. It then adds these results together for all edge points to synthesize a high-frequency image.
When the smoothing filters hm have an even number of taps and the high-pass filters gm accordingly have odd symmetry, it is advantageous to carry out the edge synthesis indirectly. One reason is that in this case, the results of filtering different standard edges by the same gm need not be the same. Another reason is that the resulting edge synthesis filters em can have an unlimited number of non-zero taps.
If the differences dem (t) between adjacent coefficient values in em are taken, however, these differences have the desirable properties of depending only on the edge sharpness, and being zero except in a limited domain. These differences are given by the following equation, in which L has the same meaning as before (L=2m-1 N). They represent the result of filtering a standard edge of unit sharpness with the high-pass filter gm, then taking differences between the resulting values at adjacent pixels. ##EQU11##
de.sub.m (t)=de.sub.m (-t)=0 if t>L
The horizontal edge image Shm is first filtered horizontally by dem (t) to obtain a horizontal differential high-frequency image Δ Yh, and the vertical edge image Svm is filtered vertically with de(t) to obtain a vertical differential high-frequency image Δ Yv. These differential high-frequency images are described by the following equations: ##EQU12##
The horizontal and vertical differential high-frequency images are then cumulatively summed to obtain the horizontal and vertical high-frequency images Yhm and Yvm. If the horizontal and vertical coordinates i and j both start at zero, these accumulation operations can be described as follows:
Yh.sub.m (0, j)=Δ Yh.sub.m (0, j)
Yh.sub.m (i, j)=Δ Yh.sub.m (i, j)+Yh.sub.m (i-1, j) (i>0)
Yv.sub.m (i, 0)=Δ Yv.sub.m (i, 0)
Yv.sub.m (i, j)=Δ Yv.sub.m (i, j)+Yv.sub.m (i, j-1) (j>0)
In the inverse wavelet transform processor 36, each inverse wavelet processor 36-m performs three two-dimensional filtering operations, using the conjugate filters f*m, g*m, and h*m of fm, gm, and hm. The image X'm is filtered by h*m horizontally and vertically, the horizontal high-frequency image Yh*m is filtered by g*m horizontally and f*m vertically, and the vertical high-frequency image Yvm is filtered by g*m vertically and f*m horizontally. The three results are added to obtain the partially reconstructed image X'm-1. This inverse wavelet transform is well known in the art, and is described by the following equation: ##EQU13##
The reconstructed image X'0 is output to the second input/output device 37 for display or print-out. Alternatively, the reconstructed image can be output to a computer for further image processing.
Synthesizing high-frequency image information from edge images as described above enables high compression ratios to be obtained. High-frequency information corresponding to edges below the sharpness threshold T is lost, so there is some loss of low-level, high-frequency shading variations, but the reconstructed image is not distorted, and sharp edges are faithfully preserved.
One feature of the invented encoder is that edge detection is performed only at the original image scale, which permits edge detection to proceed concurrently with the smoothing filtering and down-sampling operations in the encoder, to speed up the encoding process. Another feature is that in the decoder, high-frequency images are synthesized in a single step at each wavelet scale, without recourse to iterative procedures. This reduces the computational load in the decoder, speeds up the decoding process, and eliminates the prior-art problem of deciding when to stop iterating.
Second embodiment
The second embodiment is similar to the first, but incorporates a down-sampling process into each of the smoothing filters in the encoder and decoder, and an up-sampling process into each of the inverse wavelet processors in the decoder.
Referring to FIG. 4, the digital image encoder in the third embodiment has the same edge detector 21 and encoding section 27 as in the first embodiment, but its two-dimensional smoothing and down-sampling section 41 employs smoothing and down-sampling filters 42-m (m=1 to n). In addition to low-pass filtering, these cascaded filters 42-m down-sample the filtered data by discarding every other pixel in both the horizontal and vertical directions.
FIG. 5 shows a conceptual flow diagram of the smoothing and down-sampling filtering process. The first step 43 is a vertical smoothing step carried out using a low-pass filter h identical to filter h1 in the first embodiment. The second step 44 is a vertical down-sampling step that discards every other pixel in the vertical direction. The third step 45 is a horizontal smoothing step in which the remaining image data are filtered by the same filter h in the horizontal direction. The fourth step 46 is a horizontal down-sampling step that discards every other pixel in the horizontal direction.
Needless to say, the values of pixels that will be discarded immediately following a filtering operation do not have to be computed in that filtering operation. The entire operation performed by the two-dimensional smoothing and down-sampling filter 42-m can therefore be described by the following equation: ##EQU14##
This operation reduces the size of the image by a factor of two in both the horizontal and vertical directions. The same low-pass filter h is used in all the filters 42-m (m=1 to n). It is not necessary to double the tap spacing by inserting zero coefficients at successive wavelet scales as in the first embodiment.
In the drawings, the symbol hd denotes filtering by h followed by down-sampling as described above.
FIG. 6 shows the digital image decoder of the second embodiment. The decoding section 29 is the same as in the first embodiment. The output of the reduced image decoder 32 is now denoted X'n, but is similar to the output R in the first embodiment. The decoder employs filters e, f, g, and h that are identical to e1, f1, g1, and h1 in the first embodiment. The subscripts d and u attached to these filters denote associated down-sampling and up-sampling operations, respectively.
The edge images Sv1 and Sh1 output by the edge image decoder 31 are filtered by a cascade of smoothing and down-sampling filters 48-m (m=1 to n-1) identical to the smoothing and down-sampling filters 42-m in the encoder. Each filter 48-m filters and smooths each input edge image Svm and Shm in both the vertical and horizontal directions, and down-samples the result by a factor of two in both directions to generate edge images Svm+1 and Shm+1. Due to the down-sampling, Svm+1 is only half as wide and half as high as Svm, and Shm+1 is only half as wide and half as high as Shm.
The decoded edge images Svm and Shm are input to an edge synthesizer 49-m (m=1 to n). There, the vertical edge image Svm is down-sampled by a factor of two in the horizontal direction, filtered by the above-mentioned filter e in the vertical direction, then down-sampled by a factor of two in the vertical direction, while the horizontal edge image Shm is down-sampled by a factor of two in the vertical direction, filtered by filter e in the horizontal direction, then down-sampled by a factor of two in the horizontal direction. These operations are depicted conceptually in FIG. 7. The resulting high-frequency images will be denoted Yhm and Yvm as in the first embodiment.
The high-frequency images Yhm and Yvm (m=1 to n) and the low-frequency reduced image X'n are input to an inverse wavelet transform processor 50 comprising cascaded inverse wavelet processors 50-m (m=1 to n). Processors 50-m use the conjugate filters f*, g*, and h* to perform the same computation as in the first embodiment, but each filtering step in the computation is preceded by an up-sampling step. These up-sampling steps double the image size by inserting zero-valued pixels between every pair of adjacent pixels in the appropriate direction. To prevent the zero-valued pixels from reducing the mean value of the output image, after each filtering step, the resulting pixel values are doubled.
Due to up-sampling, the image X'm-1 output from inverse wavelet processor 50-m is twice as wide and twice as high as each of the three images X'm, Yvm, and Yhm input to processor 50-m. The three input images X'm, Yvm, and Yhm all have the same size.
FIG. 8 depicts the operation performed by the inverse wavelet processor 50-m schematically. As indicated at the top of this drawing, the partially reconstructed image X'm received from the preceding processor 50-(m+1) in the cascade is up-sampled horizontally by interpolating zeros (reducing the mean pixel value by a factor of two) and filtered horizontally by h*; then the resulting pixel values are doubled (restoring the correct mean value). The image is now up-sampled vertically, again by interpolating zeros, and filtered by h* vertically. Similarly, high-frequency image Yhm is up-sampled vertically, filtered by f* vertically, doubled in pixel value, up-sampled horizontally, and filtered by g* horizontally. High-frequency image Yvm is up-sampled horizontally, filtered by f* horizontally, doubled in pixel value, up-sampled vertically, and filtered by g* vertically.
The three resulting images are then added together to produce a sum image, all pixel values in which are again doubled to counteract the effect of the second up-sampling steps. The final result is the partially reconstructed image X'm-1 which is passed to the next wavelet processor 50-(m-1), or the reconstructed image X'0 output to the input/output device 37.
The operation performed by inverse wavelet processor 50-m can be described by the following equation. Up-sampling is accomplished by dividing the values of i, j, k, and l by two. If either i/2+k/2 or j/2+1/2 is not an integer, the corresponding values of X'm, Yhm, and Yvm are taken to be zero. ##EQU15##
The second embodiment provides the same advantages as the first embodiment but requires less computation, because the amount of image data to be processed is reduced by a factor of four (by a factor of two horizontally and another factor of two vertically) at each wavelet scale. The computational load of the additional up-sampling and down-sampling steps, and of multiplying pixel values by two, is slight.
Third embodiment
The third embodiment detects edges diagonally as well as horizontally and vertically. To prevent an increase in the amount of encoded data, some of the down-sampling steps that were performed in the decoder in the second embodiment are moved into the encoder.
FIG. 9 is a block diagram of the digital image encoder of the third embodiment. The two-dimensional smoothing and down-sampling section 41 is the same as in the second embodiment. The original image X0 is also input to a one-dimensional smoothing and down-sampling section 51, which outputs the original image X0 together with a vertically smoothed and down-sampled image Xh and a horizontally smoothed and down-sampled image Xv to an edge detector 52. The edge detector 52 outputs three edge images Sh, Sv, and Sc to the encoding section 27, which is the same as in the first two embodiments.
FIG. 10 illustrates the processes carried out in the one-dimensional smoothing and down-sampling section 51. Image Xv is produced by filtering the original image X0 with the low-pass filter h in the horizontal direction, then down-sampling the resulting image by removing every other pixel in the horizontal direction. Image Xv accordingly contains only half as much data as image X0.
Similarly, image Xh is produced by filtering X0 vertically with the low-pass filter h, then down-sampling the result by removing every other pixel in the vertical direction. Image Xh also contains only half as much data as image X0.
Images Xv and Xh can be described by the following equations: ##EQU16##
The edge detector 52 detects edges by the method described earlier: by taking differences between adjacent pixels, then taking differences between these differences. This method is applied in the directions that have not been down-sampled. Accordingly, differences are taken horizontally in the vertically down-sampled image Xh, producing a horizontal edge image Sh. Similarly, differences are taken vertically in the horizontally down-sampled image Xv, producing a vertical edge image Sv.
Edges are detected diagonally by processing the original image X0 in both the horizontal and vertical directions. The difference-of-differences operation is first performed on X0 in, for example, the horizontal direction, producing an intermediate image. The difference-of-differences operation is next performed on this intermediate image in the vertical direction. Resulting pixels with absolute values equal to or less than a threshold value T are then set to zero, and other values are suitably quantized, producing a diagonal edge image Sc. Prior to quantization, Sc has the following values: ##EQU17##
The edge image encoder 24 can encode the diagonal edge image Sc by methods similar to those used for the vertical and horizontal edge images Sv and Sh. Sc has the same size as the original image X0, while Sv and Sh are only half this size, so the total amount of data to be encoded is the same as in the first two embodiments.
FIG. 11 is a block diagram of the digital image decoder in the third embodiment. The decoding section 29 is the same as in the preceding embodiments, except that the edge image decoder 31 outputs a set of three edge images, Sv1, Sh1, and Sc1, corresponding to the encoded edge images Sv, Sh, and Sc.
These images are input to a cascade of two-dimensional smoothing and down-sampling filters 53-m (m=1 to n-1). These are identical to the two-dimensional smoothing and down-sampling filters 48-m in the second embodiment, except that each filter smooths and down-samples three separate edge images instead of two, producing vertical, horizontal, and diagonal edge images Svm+1, Shm+1, and Scm+1. Each of these edge images Svm+1, Shm+1, and Scm+1 is only half as high and half as wide as the corresponding image Svm, Shm, or Scm in the preceding set.
These images are processed by corresponding edge synthesizers 54-m as explained below to create sets of vertical, horizontal, and diagonal high-frequency images Yvm, Yhm, and Ycm. These are supplied to an inverse wavelet transform processor 55, comprising cascaded inverse wavelet processors 55-m (m=1 to n), which output reconstructed or partially reconstructed images X'm-1.
Edge synthesizer 54-m operates as illustrated in FIG. 12, using the same edge synthesis filter e as in the second embodiment. The vertical edge image Svm is filtered vertically by e, then down-sampled by removing every other pixel in the vertical direction to create a vertical high-frequency image Yvm. Similarly, the horizontal edge image Shm is filtered horizontally by e, then down-sampled horizontally to create a horizontal high-frequency image Yhm. The diagonal edge image Scm is filtered vertically by e, down-sampled in the vertical direction, filtered horizontally by e, then down-sampled in the horizontal direction, creating a diagonal high-frequency image Ycm.
Due to the combination of down-sampling in the one-dimensional smoothing and down-sampling section 51 in the encoder and in the edge synthesizer 54-m in the decoder, the three high-frequency images Yvm, Yhm, and Ycm are all the same size. Each has been down-sampled m times in both the horizontal and vertical directions, reducing the number of pixels by a total factor of 4m.
Inverse wavelet processor 55-m operates on images X'm, Yvm, Yhm, and Ycm as illustrated in FIG. 13, using the same complementary pair of high-pass and low-pass filters g* and h* as in the second embodiment, but not using the completion filter f*. As in the second embodiment, each filtering step is preceded by an up-sampling step and followed by a compensatory doubling of the pixel values. Up-sampling is carried out by inserting zero-valued pixels.
Inverse wavelet processor 55-m filters each of the four input images horizontally and vertically, and adds the four resulting images together to obtain the output image X'm-1. Image X'm is filtered horizontally and vertically by h*; image Yhm is filtered horizontally by g* and vertically by h*; image Yvm is filtered horizontally by h* and vertically by g*; and image Ycm is filtered horizontally and vertically by g* For computational efficiency, the results of the four horizontal filtering operations are added pairwise, so that only two vertical filtering operations have to be performed.
The operation performed by inverse wavelet processor 55-m can also be described by the following equation. As before, if either i/2+k/2 or j/2+1/2 is not an integer, the corresponding values of X'm, Yhm, Yvm, and Ycm are zero. ##EQU18##
The third embodiment provides the same advantages as the second and offers improved rendition of high-frequency detail in the reconstructed image, due to diagonal edge detection.
Fourth embodiment
The fourth embodiment is similar to the third embodiment, differing only in the operation of the edge detector 52. Block diagrams will be omitted.
The edge detector 52 of the fourth embodiment begins by detecting horizontal, vertical, and diagonal edge sharpness by the methods described in the third embodiment, using a threshold value T. It next constructs thickened edges by including contiguous points with edge sharpness values exceeding a lower threshold U. Finally, it trims the edges so that each edge has a uniform thickness (although some edges may be thicker than others). This process will be explained below for an edge in the horizontal edge image Sh.
FIG. 14 shows part of a typical horizontal edge image Sh, using stippling to denote pixels with sharpness values exceeding T, and hatching to denote pixels exceeding U but not exceeding T. Here, "exceeding" means having a positive edge sharpness value exceeding T (or U). Pixels with negative sharpness values are processed separately.
Edge points in the horizontal edge image Sh tend to be organized into chains with a generally vertical orientation. FIG. 14 shows one such chain, running generally from the top left toward the bottom right. This edge is irregular in shape, however, and if detected by threshold T, the edge is fragmented into two parts. Such irregularities and fragmentation greatly impair the efficiency which which the edge image can be encoded.
FIG. 15 shows the thickened version of this edge. The edge detector 52 constructs FIG. 15 from FIG. 14 by the following procedure.
Starting from the top left corner, the edge detector searches for a pixel in the horizontal edge image Sh in FIG. 14 exceeding threshold T. When the first such pixel is found, it is placed in the thickened edge image in FIG. 15 as the first pixel of the first layer of an edge. This first layer is then extended in the left and right directions by including any contiguous pixels having the same sign and exceeding threshold U. The extension continues as far in each direction as such contiguous pixels continue. In the drawings, pixel 56 is found first and pixels 57 and 58 are added to complete the first layer. If there were further pixels exceeding threshold U immediately to the left of pixel 57 or to the right of pixel 58, they would also be added in.
Next, if there any pixels exceeding threshold T immediately below the bottom layer of the edge so far, they are added to begin a new layer of the edge. This new layer is then extended to the left and right by adding contiguous pixels exceeding threshold U. In this extension process, all contiguous pixels exceeding threshold T are added in. Pixels exceeding threshold U but not exceeding threshold T are added in only if they are not disposed more than a certain limit extension distance beyond the ends of the preceding layer. In this example, the limit extension distance is one pixel. Thus pixel 59 is added, because it extends only one pixel past pixel 57, but the pixels to the left of pixel 59 would not be added to this layer unless they exceeded threshold T.
The edge detector continues in this manner, extending the thickened edge downward by adding new layers as long as it can find a pixel exceeding threshold T immediately below the bottom layer so far, and enlarging each layer horizontally to include contiguous pixels exceeding threshold T, as well as pixels exceeding threshold U out to the limit extension distance. In this example, pixel 60 is not added to the thickened edge because it would extend two pixels beyond the preceding layer, and the extension limit is one pixel.
When no pixels exceeding threshold T can be found below the bottom layer, the thickened edge is considered complete. The edge detector then searches for another pixel exceeding threshold T in absolute value to start a new thickened edge, and thickens this edge in the same way. This entire procedure is repeated until no more pixels exceeding threshold T can be found. A similar procedure is then followed to process pixels having negative sharpness values exceeding -T and -U.
Although the limit extension distance in the example above was one pixel, other limit extension distances such as zero, two, or three pixels can be used. The preferable limit extension distance depends in part on the threshold values T and U.
FIG. 16 shows the trimmed version of the thickened edge in FIG. 15. The trimmed edge is constructed by comparing adjacent layers in the thickened edge, and trimming the longer layer to match the length of the shorter layer. Thus pixel 59 is trimmed to reduce the length of its layer to three pixels, matching the length of the layer above. This trimming procedure is carried out from the top of the edge to the bottom, then repeated until all layers in the edge have been trimmed to the same length.
If the longer layer extends beyond the shorter layer at only one end, the extending end is trimmed. If the longer layer extends beyond the shorter layer at both ends, the edge detector 52 examines the preceding layer, and trims in such a way as to maintain a constant offset from one layer to the next. In FIG. 16, for example, after trimming layers 61 and 62, the edge detector 52 trims layer 63 so that it is offset one pixel to the right of layer 62, because layer 62 is similarly offset one pixel to the right of layer 61.
The trimmed edge in FIG. 16 is continuous, and has a more regular shape than the original edge in FIG. 14. These properties enable the edge in FIG. 16 to be encoded more efficiently than the edge in FIG. 14, even though it contains more pixels. In chain encoding, for example, after the position and length of the first layer have been specified, the position of the rest of the edge can be specified just by giving the offset of each layer from the preceding layer. In quantization of the edge sharpness values, pixels exceeding threshold U but not threshold T can be quantized to zero but then dequantized to a non-zero value in the digital image decoder.
In the vertical edge image Sv, edges are thickened and trimmed by a similar procedure, except that the thickening and trimming are done vertically, and the edge detector 52 works from left to right instead of from top to bottom.
Edges in the diagonal edge image Sc can be processed either horizontally or vertically. The choice depends on the way in which the diagonal edge image will be encoded. If vertical chain encoding is used, for example, the thickening and trimming should be done horizontally.
Edge thickening and trimming can also be added to the edge detection process in the first and second embodiments, and in any of the embodiments that follow. The invented edge thickening and trimming technique is moreover useful for regularizing edges in image encoding and image processing in general, and is not limited to edges detected by the difference-of-differences method.
Fifth embodiment
The fifth embodiment is similar to the second embodiment, but is adapted to process moving images comprising successive frames.
Referring to FIG. 17, the digital image encoder of the fifth embodiment has the same two-dimensional smoothing and down-sampling section 41 and edge detector 21 as in the second embodiment, but a different encoding section 65. The encoding section 65 comprises a pair of motion estimators 66 and 67, a motion vector encoder 68, a pair of motion compensators 69 and 70, and a pair of memories 71 and 72. It also has the same edge image encoder 24 and reduced image encoder 25 as in the digital image encoders of the preceding embodiments, and the same edge image decoder 31 and reduced image decoder 32 as in the digital image decoders of the preceding embodiments.
Motion estimator 66 receives the pair of edge images Sh and Sv from the edge detector 21. To simplify the drawing, this pair of edge images is denoted by the single letter S. Motion estimator 66 also receives a pair of previous edge images S" (a previous horizontal edge image and a previous vertical edge image) from memory 71. Motion estimator 66 outputs edge motion vectors Ms and a pair of horizontal and vertical edge difference images, denoted for simplicity by the single symbol Ds.
Motion estimator 67 receives the reduced image Xn from the last smoothing filter 42-n and a previous reduced image X"n from memory 72, and outputs reduced motion vectors Mr and a reduced difference image Dr. The motion vector encoder 68 encodes motion vectors Ms and Mr and outputs motion information Cm.
The edge image encoder 24 encodes the edge difference images Ds and outputs edge image information Cs. This information is decoded by the edge image decoder 31 to produce a pair of decoded edge difference images, which are denoted in the drawing by the single symbol D's. Motion compensator 69 receives this pair of images D's, the previous pair of images S", and the motion vectors Ms, and produces a pair of decoded edge images S1, which it stores in memory 71.
Similarly, the reduced image encoder 25 encodes the reduced difference image Dr and outputs reduced image information Cr. The reduced image decoder 32 decodes Cr and outputs a decoded reduced difference image D'r. Motion compensator 70 receives this image D'r, the previous reduced image X"n, and the motion vectors Mr, and produces a decoded reduced image X'n which it stores in memory 72.
The multiplexer 73 multiplexes the edge image information Cs, motion information Cm, and reduced image information Cr for output as an encoded image C.
Referring to FIG. 18, the digital image decoder of the fifth embodiment has a decoding section 75 comprising a demultiplexer 76, a motion vector decoder 77, and decoders 31 and 32, motion compensators 69 and 70, and memories 71 and 72 identical to those in the digital image encoder in FIG. 17. The rest of the digital image decoder in FIG. 18 is the same as in the second embodiment, comprising smoothing and down-sampling filters 48-m (m=1 to n-1), edge synthesizers 49-m (m=1 to n), and an inverse wavelet transform processor 50.
The demultiplexer 76 receives the encoded image C from the input/output device 28 and demultiplexes it into edge image information Cs, motion information Cm, and reduced image information Cr. The motion vector decoder 77 decodes Cm to obtain decoded edge motion vectors M's and reduced motion vectors M'r. Other symbols in FIG. 18 are as already explained in FIGS. 6 and 17, with Sm representing the pair of edge images Shm and Svm, and Ym the pair of high-frequency images Yhm and Yvm.
Next the operation of the fifth embodiment will be described. The description will be confined mainly to the encoding section 65 of the encoder.
Referring again to FIG. 17, motion estimator 67 estimates motion in the reduced image Xn by, for example, the well-known block-matching method. This method involves comparing a block of pixels in the image Xn with blocks displaced by k pixels horizontally and l pixels vertically from this block in the previous image X"n in memory 72, and selecting the vector (k, l) that gives the closest match. These vectors (k, l) are output as motion vectors Mr, while the differences between the pixel values in Xn and the pixels values in X"n at positions offset by these motion vectors (k, l) are output as the reduced difference image Dr. Thus the reduced difference image Dr is the difference between Xn and X"n over and above the motion represented by the motion vectors Mr.
Motion estimator 66 carries out a similar process separately on the horizontal and vertical edge images to obtain the motion vectors Ms and the pair of edge difference images denoted Ds. The motion vector encoder 68 encodes the motion vectors Ms and Mr by a lossless encoding method such as a variable-length encoding method.
The encoders 24 and 25 encode the difference images Ds and Dr by, for example, the same methods as used in the preceding embodiments. The decoders 31 and 32 employ methods reverse to those methods to obtain the decoded difference images D'r and D's.
Motion compensator 70 adds the decoded reduced difference image D'r to the previous reduced image X"n according to the motion vectors Mr, and stores the resulting image X'n in memory 72. "According to the motion vectors Mr" means that each pixel value in D'r is added to a corresponding pixel value in the previous reduced image X"n at a location offset by a motion vector (k, l) in Mr. The method by which X'n is obtained from D'r is inverse to the method by which Dr is obtained from Xn, and the operations of encoder 25 and decoder 32 are mutually inverse, so image X'n is substantially identical to image Xn.
Similarly, motion compensator 69 adds the two decoded edge difference images to the corresponding previous edge images according to the motion vectors Ms, and stores the resulting pair of decoded edge images S1 in memory 71. The pair of edge images S1 are substantially identical to the pair of edge images S output by the edge detector 21.
Memory 72 has space for storing at least two complete reduced images: the previous image X"n supplied to motion estimator 67 and motion compensator 70, and the new image X'n stored by motion compensator 70. At the end of each frame, the previous image X"n is discarded, leaving the new image X'n to be used as the previous image in the next frame. Memory 71 stores edge images in a similar fashion.
From time to time the motion estimators 66 and 67 may ignore the previous images S" and X"n in the memories 71 and 72 and supply the edge images S and reduced image Xn directly to the encoders 24 and 25, instead of supplying difference images. The motion information Cm in this case consists of a message indicating that the current frame does not require motion compensation. The motion compensators 69 and 70 then simply output the images received from the decoders 31 and 32.
The motion compensators 69 and 70 and memories 71 and 72 in the digital image decoder operate in the same way as the motion compensators 69 and 70 and memories 71 and 72 in the digital image encoder, using the decoded motion vectors M's and M'r supplied by the motion vector decoder 77. Since a lossless encoding scheme is employed for the motion vectors, M's and M'r are exactly identical to Ms and Mr, and the images denoted S1 and X'n in FIG. 18 are exactly identical to the images denoted S1 and X'n in FIG. 17. Further operation of the digital image decoder is as already described in the second embodiment.
By removing temporal redundancy, motion compensation can dramatically increase the coding efficiency of moving images, but block matching conventionally requires considerable computation. One advantage of the fifth embodiment is that most of the block matching is performed in the edge images, which have been thresholded and tend to consist mostly of zero-valued pixels. The large number of zero-valued pixels greatly speeds up the block-matching process. Further gains in speed are possible by using motion vectors calculated for one of the two edge images to help find motion vectors in the other edge image.
Another advantage is that the motion estimators and compensators in the digital image encoder and digital image decoder use identical previous images S" and X"n. This prevents the accumulation of quantization error and other image errors resulting from compressive coding in the encoders 24 and 25, and ensures consistency between motion estimation in the digital image encoder and motion compensation in the digital image decoder.
Other advantages of the fifth embodiment are as described in the second embodiment. The fifth embodiment can be employed in video teleconferencing and a wide variety of other multimedia applications.
Sixth embodiment
The sixth embodiment is similar to the fifth embodiment, with further provisions for speeding up the computation of motion vectors.
Referring to FIG. 19, the encoding section 78 of the digital image encoder in the sixth embodiment is similar to the encoding section 65 in the fifth embodiment, but has a bidirectional interconnection 79 between motion estimators 66 and 67. This enables motion estimator 67 to receive the edge motion vectors Ms found by motion estimator 66, scale them down to the size of the low-frequency reduced image Xn, and restrict the search for motion vectors in image Xn to vectors close to the scaled-down edge motion vectors. In the extreme case, the search can be eliminated entirely: the reduced motion vectors Mr can be computed directly from the edge motion vectors Ms.
Alternatively, motion estimator 66 can receive the motion vectors Mr found by block matching in the low-frequency reduced image Xn, scale them up to the size of the edge images Sh and Sv, and restrict the search for motion vectors in Sh and Sv to vectors close to the scaled-up vectors. This scheme is particularly effective in detecting large, rapid motion.
Aside from this additional interconnection 79, the sixth embodiment has the same structure as the fifth embodiment and operates in the same way, so further description will be omitted. By narrowing down the search for motion vectors, the sixth embodiment enables frames to be encoded at a faster rate than in the fifth embodiment.
Seventh embodiment
The seventh embodiment combines features of the third and fifth embodiments by detecting edges horizontally, vertically, and diagonally, and employing motion compensation.
Referring to FIG. 20, the digital image encoder of the seventh embodiment has the same two-dimensional smoothing and down-sampling section 41, one-dimensional smoothing and down-sampling section 51, and edge detector 52 as in the third embodiment (FIG. 8), and the same encoding section 65 as in the fifth embodiment (FIG. 17). The symbol S in FIG. 20 denotes the set of three edge images Sv, Sh, and Sc in FIG. 8. Each of the symbols S1, S", Ds, and D's likewise denotes a set of three edge images.
Referring to FIG. 21, the digital image decoder of the seventh embodiment has the same decoding section 75 as the fifth embodiment (FIG. 17), and the same two-dimensional smoothing and down-sampling filters 53-1 to 53-(n-1), edge synthesizers 54-1 to 54-n, and inverse wavelet transform processor 55 as in the third embodiment (FIG. 11).
The seventh embodiment operates as explained in the third and fifth embodiments, and provides similar effects, so further description will be omitted.
Eighth embodiment
The eighth embodiment combines features of the sixth and seventh embodiments.
Referring to FIG. 22, the digital image encoder of the eighth embodiment has the same configuration as in the seventh embodiment (FIG. 20), except for the addition of a bidirectional interconnection 79 between motion estimators 66 and 67 as in the sixth embodiment (FIG. 19). The eighth embodiment operates as explained in the sixth and seventh embodiments, so further description will be omitted.
Ninth embodiment
The ninth embodiment performs motion estimation before edge detection, smoothing, and down sampling. To maintain consistency between the encoder and decoder, the encoder performs substantially all the processes carried out in the decoder, including edge synthesis and the inverse wavelet transform.
Referring to FIG. 23, the original image frames input at the input terminal 10 are supplied to a motion estimator 80, which performs motion estimation as described above, outputting a difference image Do and motion vectors M. The motion vectors M are encoded by a motion vector encoder 68 as in the fifth to eighth embodiments. The difference image D0 is supplied to an edge detector 21 and two-dimensional smoothing and down-sampling section 41 in place of the original image X0, and is processed in the same way that the original image X0 was processed in the second embodiment. The resulting pair of edge images will again be denoted by the symbol S, while the reduced image resulting from smoothing and down-sampling of D0 will be denoted Dn. These images S and Dn are encoded by respective encoders 24 and 25 to produce encoded image information Cs and Cr, which is combined with the motion information Cm by a multiplexer 73 as in the preceding embodiments.
The encoded information Cs and Cr is also decoded by decoders 31 and 32 to produce decoded images S1 and D'n. These are supplied to two-dimensional smoothing and down-sampling filters 48-m (m=1 to n-1), edge synthesizers 49-m (m=1 to n), and an inverse wavelet transform processor 50 which are identical to the corresponding elements in the digital image decoder of the second embodiment, shown in FIG. 6. The inverse wavelet transform processor 50 outputs a reconstructed difference image D'0 to a motion compensator 81.
Using this image D'0, the motion vectors M output by the motion estimator 80, and a previous image X"0 stored in a memory 82, the motion compensator 81 produces a new reconstructed image X'0, which is stored in the memory 82. The previous image X"0 was also supplied to the motion estimator 80 for use in motion estimation. The reconstructed image X'0 will become the previous image in the next frame.
The motion estimator 80, motion compensator 81, and memory 82 in FIG. 23 operate like the motion estimator 67, motion compensator 70, and memory 72 in the fifth or seventh embodiment (FIG. 17 or 20), although they operate on the original image X0 instead of the reduced image Xn. The inverse waveform processors 50-m in FIG. 23 operate like the inverse waveform processors 50-m in the second embodiment (FIG. 6), although they process a series of partially reconstructed difference images D'm instead of a series of partially reconstructed images X'm (m=1 to n), and output a fully reconstructed difference image D'0 instead of a fully reconstructed image X'0.
The encoding section 83 in the ninth embodiment comprises the motion vector encoder 68, edge image encoder 24, reduced image encoder 25, and multiplexer 73.
Referring to FIG. 24, the digital image decoder of the ninth embodiment has a decoding section 84 with a demultiplexer 76 and motion vector decoder 77 similar to those in the fifth to eighth embodiments. The other elements in FIG. 24 are similar to the elements with the same reference numerals in FIG. 23. The motion compensator 81 receives a fully reconstructed difference image D'0 from the inverse wavelet transform processor 50, and outputs a fully reconstructed image X'0 to the input/output device 37.
As in the fifth to eighth embodiments, the digital image encoder in FIG. 23 and digital image decoder in FIG. 24 operate with identical previous images X"0, thereby maintaining consistency between motion estimation in the encoder and decoder. Performing motion estimation before edge detection can lead to higher compression ratios than in the fifth to eighth embodiments, because only one set of motion vectors has to be encoded.
Tenth embodiment
The tenth embodiment combines features of the third and ninth embodiments. Three edge images are detected, as in the third embodiment. Edge detection, smoothing, and down-sampling are preceded by motion estimation, and followed by all the processes performed in the digital image decoder, as in the ninth embodiment.
Referring to FIG. 25, the digital image encoder of the tenth embodiment has the motion estimator 80, motion compensator 81, memory 82, and encoding section 83 of the ninth embodiment (FIG. 23), and other elements as in the third embodiment (FIGS. 9 and 11). Referring to FIG. 26, the digital image decoder has the decoding section 84, motion compensator 81, and memory 82 of the ninth embodiment (FIG. 24), and other elements as in the third embodiment (FIG. 11).
The tenth embodiment operates as described in the third and ninth embodiments, and provides similar advantages. A detailed description will be omitted.
Eleventh embodiment
The eleventh embodiment is a digital image encoder/decoder (codec) that combines the functions of the digital image encoder and decoder of the ninth embodiment. The eleventh embodiment thus provides one device that can operate in either an encoding mode or a decoding mode.
Referring to FIG. 27, to the digital image encoder of the ninth embodiment (FIG. 23), the eleventh embodiment adds the demultiplexer 76 and motion vector decoder 77 of the digital image decoder of the ninth embodiment (FIG. 24), a second input terminal 90, and three switches 91, 92, and 93.
To encode a moving image, switches 91, 92, and 93 are set to the position marked "a" in the drawing. Successive frames X0 are input at input terminal 10 and processed just as in the digital image encoder of the ninth embodiment, the encoded results being output from the demultiplexer 73 to input/output device 28.
To decode an encoded moving image, switches 91, 92, and 93 are set to the position marked "b." The encoded moving image C is input from the second input terminal 90 to the demultiplexer 76 and processed just as in the digital image decoder of the ninth embodiment, the decoded results being output from the motion compensator 81 to input/output device 37.
The eleventh embodiment has the same advantages as the ninth embodiment, with the further advantage of allowing the encoding and decoding functions to share the same motion compensator 81, decoders 31 and 32, filters 48-m, edge synthesizers 49-m, and inverse wavelet transform processor 50, thereby avoiding needless duplication. This embodiment is particularly suited for videophones and other bidirectional video systems.
Twelfth embodiment
The twelfth embodiment is a digital image encoder/decoder that combines the features of the tenth and eleventh embodiments.
Referring to FIG. 28, the twelfth embodiment adds to the digital image encoder of the tenth embodiment (FIG. 25) the same demultiplexer 76, motion vector decoder 77, second input terminal 90, and switches 91, 92, and 93 as in the eleventh embodiment. The twelfth embodiment operates like the eleventh, but has the further advantage of improved rendition of high-frequency image detail due to diagonal edge detection.
As described above, the invented encoding and decoding method and devices provide high compression rates with modest amounts of computation. Although there is some loss of detail, reconstructed images are not distorted, and because high-frequency information is synthesized in a single step instead of by iteration, the reconstruction process is reliably completed within a short and definite time. The invention can be usefully applied to both still and moving images for a wide variety of purposes, in a wide variety of systems.
The scope of the invention is not restricted to the embodiments described above, but includes such further variations as the following.
In the first, second, fifth, sixth, ninth, and eleventh embodiments, instead of producing separate horizontal and vertical edge images, the edge detector 21 can output a single edge image in which each edge point has both a sharpness value S(i, j) and direction angle θ (i, j), computed from Sh(i, j) and Sv(i, j) by, for example, the following formulas:
S(i, j)=sgn Sh(i, j)!· Sh(i, j).sup.2 +Sv(i, j).sup.2 !.sup.1/2
θ(i, j)=arctan Sv(i, j)/Sh(i, j)!
In these formulas, sgn denotes the algebraic sign (plus or minus one), and arctan the inverse tangent function. The encoding section then has only one set of edge positions to encode. The decoding section in the digital image decoder can recover Sh(i, j) and Sv(i, j) from S(i, j) and θ (i, j).
This method of encoding only a single edge image is particularly effective if chain encoding is employed, because the direction angle tends to change only slowly along a chain of edge points. Moreover, the angle tends to be perpendicular to the direction of the chain, so it is possible to have the edge detector calculate only the above edge sharpness S(i, j), and let the decoder infer the direction angle θ (i, j) of an edge point (i, j) from the direction of the chain in the vicinity of that point. The compression ratio can be significantly improved in this way, although with some loss of reconstruction accuracy.
This encoding method can also be applied to the horizontal and vertical edge images in the other embodiments, if the original image is not down-sampled before horizontal and vertical edge detection. The necessary down-sampling is then performed instead in the decoder, as in the second embodiment. In the fifth, sixth, seventh, and eighth embodiments, this encoding method has the further advantage of reducing the number of edge images for which motion vectors have to be computed.
In the digital image encoder in any of the embodiments, it is possible to have the two-dimensional smoothing performed in a single step, instead of by a cascaded series of filters. The filters h1, h2, . . . , hn in the first embodiment, for example, can be mutually convolved to yield a single filter that gives the same result as applying the cascaded series of filters. For use in the other embodiments, down-sampling can be incorporated into this single filter simply by calculating the values of only 1/4n of the output pixels.
The smoothing filters applied in the horizontal and vertical directions need not be identical. A smoothing filter with an even number of taps can be applied in one of these directions, and a smoothing filter with an odd number of taps in the other direction. The edge synthesis filters (e), completion filters (f), and high-pass filters (g) employed for decoding then also require separate versions for the horizontal and vertical directions.
Edge detection is not limited to the difference-of-differences method. This method is preferred, but other methods can be employed.
The filters employed were described as being symmetric around zero or 1/2, but this is only a matter of notation. The center of symmetry can be disposed at any convenient location.
The standard edge from which the edge synthesis filters are generated need not be an edge of the type shown in FIG. 3, in which pixel values change at first one constant rate, then another constant rate. Other types can be used, provided the standard edge has only one point which would be detected as an edge point by the edge detector.
Those skilled in the art will recognize that still further variations are possible without departing from the scope claimed below.

Claims (7)

What is claimed is:
1. A method of constructing an edge in an image consisting of pixels having sharpness values designating edge sharpness, comprising the steps of:
finding a pixel having a sharpness value exceeding a first threshold;
designating said pixel as a first pixel of a first layer of said edge;
extending said first layer in first and second directions to include contiguous pixels having sharpness values exceeding a second threshold, said second threshold being less than said first threshold, and said first and said second directions being mutually opposite;
extending said edge in a third direction, perpendicular to said first and second directions, to include adjacent pixels having sharpness values exceeding said first threshold, thus adding a new layer to said edge;
enlarging the new layer in said first and said second directions by adding contiguous pixels having sharpness values exceeding said second threshold;
repeating said steps of extending said edge and enlarging the new layer thus obtained until no more adjacent pixels having sharpness values exceeding said first threshold can be found in said third direction; and
trimming all layers of said edge to a uniform width.
2. The method of claim 1, wherein said first and second directions are horizontal and said third direction is vertical.
3. The method of claim 1, wherein said first and second directions are vertical and said third direction is horizontal.
4. The method of claim 1, wherein said step of enlarging the new layer comprises:
adding contiguous pixels that have sharpness values exceeding said first threshold; and
adding contiguous pixels that have sharpness values exceeding said second threshold and are disposed within a certain limit extension distance of a preceding layer of said edge.
5. The method of claim 4, wherein said limit extension distance does not exceed three pixels.
6. The method of claim 1, wherein said step of trimming comprises the further steps of:
finding of a pair of mutually adjacent layers of said edge, said pair consisting of a longer layer and a shorter layer, said longer layer being longer than said shorter layer;
trimming said longer layer to make said pair of mutually adjacent layers equal in length; and
repeating said steps of finding a pair of mutually adjacent layers and trimming said longer layer until all layers in said edge are equal in length.
7. The method of claim 6, wherein said step of trimming said longer layer comprises:
if said longer layer extends beyond said shorter layer at only one end, trimming said one end; and
if said longer layer extends beyond said shorter layer at both ends, trimming said longer layer with reference to another layer adjacent to said pair of mutually adjacent layers, so as to maintain a constant offset between adjacent layers.
US09/069,870 1994-10-28 1998-04-30 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform Expired - Lifetime US5878172A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US09/069,870 US5878172A (en) 1994-10-28 1998-04-30 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Applications Claiming Priority (8)

Application Number Priority Date Filing Date Title
JP6264836A JPH08123967A (en) 1994-10-28 1994-10-28 Edge detecting device for image
JP6-264836 1994-10-28
JP6-264835 1994-10-28
JP26483594 1994-10-28
JP15014295A JP3308767B2 (en) 1994-10-28 1995-06-16 Image encoding / decoding device, image decoding device
JP7-150142 1995-06-16
US08/548,943 US5761341A (en) 1994-10-28 1995-10-26 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US09/069,870 US5878172A (en) 1994-10-28 1998-04-30 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
US08/548,943 Division US5761341A (en) 1994-10-28 1995-10-26 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Publications (1)

Publication Number Publication Date
US5878172A true US5878172A (en) 1999-03-02

Family

ID=27319874

Family Applications (5)

Application Number Title Priority Date Filing Date
US08/548,943 Expired - Lifetime US5761341A (en) 1994-10-28 1995-10-26 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US09/069,870 Expired - Lifetime US5878172A (en) 1994-10-28 1998-04-30 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US09/081,104 Expired - Lifetime US5991448A (en) 1994-10-28 1998-05-19 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US09/081,780 Expired - Lifetime US5949910A (en) 1994-10-28 1998-05-20 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US09/082,729 Expired - Lifetime US6101277A (en) 1994-10-28 1998-05-21 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Family Applications Before (1)

Application Number Title Priority Date Filing Date
US08/548,943 Expired - Lifetime US5761341A (en) 1994-10-28 1995-10-26 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Family Applications After (3)

Application Number Title Priority Date Filing Date
US09/081,104 Expired - Lifetime US5991448A (en) 1994-10-28 1998-05-19 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US09/081,780 Expired - Lifetime US5949910A (en) 1994-10-28 1998-05-20 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US09/082,729 Expired - Lifetime US6101277A (en) 1994-10-28 1998-05-21 Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Country Status (4)

Country Link
US (5) US5761341A (en)
EP (1) EP0709809B1 (en)
KR (1) KR100388377B1 (en)
DE (1) DE69525127T2 (en)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6018368A (en) * 1997-07-11 2000-01-25 Samsung Electro-Mechanics Co., Ltd. Scalable encoding apparatus and method with improved function of scaling motion vector
US6101277A (en) * 1994-10-28 2000-08-08 Oki Electric Industry Co., Ltd. Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US6148115A (en) * 1996-11-08 2000-11-14 Sony Corporation Image processing apparatus and image processing method
US20020016913A1 (en) * 2000-08-04 2002-02-07 Wheeler Lynn Henry Modifying message data and generating random number digital signature within computer chip
US20020026575A1 (en) * 1998-11-09 2002-02-28 Wheeler Lynn Henry Account-based digital signature (ABDS) system
US20020032860A1 (en) * 1998-11-09 2002-03-14 Wheeler Anne Mcafee Account authority digital signature
US20030014372A1 (en) * 2000-08-04 2003-01-16 Wheeler Lynn Henry Trusted authentication digital signature (tads) system
US6898319B1 (en) * 1998-09-11 2005-05-24 Intel Corporation Method and system for video frame enhancement using edge detection
US20130307929A1 (en) * 2011-02-18 2013-11-21 Sony Corporation Image processing device and image processing method
WO2018005845A1 (en) * 2016-06-30 2018-01-04 Sony Interactive Entertainment Inc. Encoding/decoding digital frames by down-sampling/up-sampling with enhancement information

Families Citing this family (56)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0679032B1 (en) * 1994-04-20 2010-06-23 Oki Electric Industry Co., Ltd. Image encoding and decoding method and apparatus using edge systhesis and inverse wavelet transform
US5970173A (en) * 1995-10-05 1999-10-19 Microsoft Corporation Image compression and affine transformation for image motion compensation
US6037988A (en) * 1996-03-22 2000-03-14 Microsoft Corp Method for generating sprites for object-based coding sytems using masks and rounding average
EP0817494A3 (en) * 1996-06-28 1998-07-22 Oki Electric Industry Co., Ltd. Image coding method and apparatus
JP4034380B2 (en) * 1996-10-31 2008-01-16 株式会社東芝 Image encoding / decoding method and apparatus
US6259738B1 (en) * 1996-10-31 2001-07-10 Kabushiki Kaisha Toshiba Video encoding apparatus and video decoding apparatus
US6073849A (en) * 1996-11-01 2000-06-13 Psc Scanning, Inc. Electronic edge detection system using a second derivative signal processor
JPH10271497A (en) * 1997-03-24 1998-10-09 Minolta Co Ltd Image processor
US6125211A (en) * 1997-09-16 2000-09-26 Lu; Chung-Ya Progressive image transmission
US6272256B1 (en) * 1997-11-26 2001-08-07 International Business Machines Corporation Fast compression of periodic halftoned bitonal images
JP3433086B2 (en) * 1998-01-22 2003-08-04 松下電器産業株式会社 Image conversion method and image conversion device
US6400831B2 (en) 1998-04-02 2002-06-04 Microsoft Corporation Semantic video object segmentation and tracking
US6711278B1 (en) 1998-09-10 2004-03-23 Microsoft Corporation Tracking semantic objects in vector image sequences
US6147699A (en) * 1998-11-10 2000-11-14 Lexmark International, Inc. Low electromagnetic emissions and improved signal quality video drive architecture for laser printers
US6236764B1 (en) * 1998-11-30 2001-05-22 Equator Technologies, Inc. Image processing circuit and method for reducing a difference between pixel values across an image boundary
US6771827B1 (en) * 1998-12-23 2004-08-03 Xerox Corporation System and method for directed acuity segmentation resolution compression and decompression
JP3206584B2 (en) * 1999-02-05 2001-09-10 日本電気株式会社 Image data processing method and apparatus
AU4052300A (en) * 1999-04-01 2000-10-23 Chein-I Chang Method and apparatus for restoration of low resolution images
US6636633B2 (en) 1999-05-03 2003-10-21 Intel Corporation Rendering of photorealistic computer graphics images
US7170941B2 (en) * 1999-08-13 2007-01-30 Patapsco Designs Inc. Temporal compression
EP1136948A1 (en) * 2000-03-21 2001-09-26 European Community Method of multitime filtering coherent-sensor detected images
US6456663B1 (en) * 2000-03-29 2002-09-24 Matsushita Electric Industrial Co., Ltd. DCT domain down conversion system that compensates for IDCT mismatch
US6611627B1 (en) * 2000-04-24 2003-08-26 Eastman Kodak Company Digital image processing method for edge shaping
AU2001270046A1 (en) * 2000-06-22 2002-01-02 The Arizona Board Of Regents On Behalf Of The University Of Arizona Agonist versus antagonist binding to g protein-coupled receptors
KR100362355B1 (en) * 2000-09-29 2002-11-23 박성모 Processor for discrete wavelet transform
US7558310B1 (en) * 2001-01-09 2009-07-07 Urbain Alfred von der Embse Multi-scale code division frequency/wavelet multiple access
US7248257B2 (en) * 2001-02-14 2007-07-24 Technion Research & Development Foundation Ltd. Low bandwidth transmission of 3D graphical data
GB0108080D0 (en) * 2001-03-30 2001-05-23 Univ Bath Audio compression
JP4316170B2 (en) * 2001-09-05 2009-08-19 富士フイルム株式会社 Image data creation method and apparatus
US7151844B2 (en) * 2001-12-06 2006-12-19 General Motors Corporation Image sensor method and apparatus having hardware implemented edge detection processing
US7630569B2 (en) * 2002-02-26 2009-12-08 Decegama Angel Real-time software video/audio transmission and display with content protection against camcorder piracy
JP4169522B2 (en) * 2002-03-22 2008-10-22 株式会社リコー Image processing apparatus, image processing program, and storage medium for storing the program
US7266247B2 (en) 2002-09-30 2007-09-04 Samsung Electronics Co., Ltd. Image coding method and apparatus using spatial predictive coding of chrominance and image decoding method and apparatus
US6757434B2 (en) * 2002-11-12 2004-06-29 Nokia Corporation Region-of-interest tracking method and device for wavelet-based video coding
GB0227743D0 (en) * 2002-11-28 2003-01-08 Rolls Royce Plc Wavelet compression
US7463688B2 (en) * 2003-01-16 2008-12-09 Samsung Electronics Co., Ltd. Methods and apparatus for removing blocking artifacts of MPEG signals in real-time video reception
US7162076B2 (en) * 2003-02-11 2007-01-09 New Jersey Institute Of Technology Face detection method and apparatus
JP2007527567A (en) * 2003-07-02 2007-09-27 セラーテム・テクノロジー・インコーポレイテッド Image sharpening with region edge sharpness correction
NO319660B1 (en) * 2003-11-17 2005-09-05 Tandberg Telecom As Procedure for interpolating pixel values
JP2005275900A (en) * 2004-03-25 2005-10-06 Noritsu Koki Co Ltd Image processing method for particulate noise suppression, program and image processor executing this method
US7653255B2 (en) 2004-06-02 2010-01-26 Adobe Systems Incorporated Image region of interest encoding
US7574059B2 (en) * 2004-10-29 2009-08-11 Broadcom Corporation System, method, and apparatus for providing massively scaled down video using iconification
US8363730B2 (en) * 2004-12-16 2013-01-29 Intel Corporation Local macroblock information buffer
JP4626473B2 (en) * 2005-10-12 2011-02-09 セイコーエプソン株式会社 Image processing apparatus, image output apparatus and method
US7518618B2 (en) * 2005-12-23 2009-04-14 Xerox Corporation Anti-aliased tagging using look-up table edge pixel identification
US20070269120A1 (en) * 2006-05-17 2007-11-22 Motorola, Inc. Video image compression using model plus difference image
US8116581B2 (en) * 2007-06-28 2012-02-14 Microsoft Corporation Efficient image representation by edges and low-resolution signal
US8208543B2 (en) * 2008-05-19 2012-06-26 Microsoft Corporation Quantization and differential coding of alpha image data
US20100014777A1 (en) * 2008-07-19 2010-01-21 Headplay (Barbados) Inc. System and method for improving the quality of compressed video signals by smoothing the entire frame and overlaying preserved detail
JP5241423B2 (en) * 2008-10-16 2013-07-17 株式会社キーエンス Image data reduction rate determination method in image processing, pattern model positioning method in image processing, pattern model creation method in image processing, image processing apparatus, image processing program, and computer-readable recording medium
KR101710883B1 (en) * 2009-11-04 2017-02-28 삼성전자주식회사 Apparatus and method for compressing and restoration image using filter information
EP2524505B1 (en) 2010-01-15 2015-11-25 Dolby Laboratories Licensing Corporation Edge enhancement for temporal scaling with metadata
JP2012216109A (en) * 2011-04-01 2012-11-08 Sony Corp Image processor and image processing method
TWI595770B (en) * 2011-09-29 2017-08-11 杜比實驗室特許公司 Frame-compatible full-resolution stereoscopic 3d video delivery with symmetric picture resolution and quality
KR101538010B1 (en) * 2012-08-31 2015-07-22 주식회사 에이투텍 Apparatus and Method for Encoding Image, Apparatus and Method for Decoding Image
KR102272108B1 (en) * 2015-02-27 2021-07-05 삼성전자주식회사 Image processing apparatus and method

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4685143A (en) * 1985-03-21 1987-08-04 Texas Instruments Incorporated Method and apparatus for detecting edge spectral features
US5420636A (en) * 1992-05-19 1995-05-30 Sony Corporation Apparatus for and method of transmitting video signal
EP0679032A2 (en) * 1994-04-20 1995-10-25 Oki Electric Industry Co., Ltd. Image encoding and decoding method and apparatus using edge systhesis and inverse wavelet transform
US5481269A (en) * 1994-05-27 1996-01-02 Westinghouse Electric Corp. General frame wavelet classifier
US5546477A (en) * 1993-03-30 1996-08-13 Klics, Inc. Data compression and decompression
US5561724A (en) * 1993-06-11 1996-10-01 Hitachi Medical Corporation Method conducting smoothing processing in a linear direction having a minimized magnitude of density change
US5598481A (en) * 1994-04-29 1997-01-28 Arch Development Corporation Computer-aided method for image feature analysis and diagnosis in mammography
US5661822A (en) * 1993-03-30 1997-08-26 Klics, Ltd. Data compression and decompression
US5712927A (en) * 1994-07-14 1998-01-27 Samsung Electronics Co., Ltd. Method and apparatus for binary-encoding image data using error diffusion with edge enhancement
US5761341A (en) * 1994-10-28 1998-06-02 Oki Electric Industry Co., Ltd. Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3182015B2 (en) * 1993-01-27 2001-07-03 テキサス インスツルメンツ インコーポレイテツド Optical image synthesis method
US5495292A (en) * 1993-09-03 1996-02-27 Gte Laboratories Incorporated Inter-frame wavelet transform coder for color video compression

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4685143A (en) * 1985-03-21 1987-08-04 Texas Instruments Incorporated Method and apparatus for detecting edge spectral features
US5420636A (en) * 1992-05-19 1995-05-30 Sony Corporation Apparatus for and method of transmitting video signal
US5546477A (en) * 1993-03-30 1996-08-13 Klics, Inc. Data compression and decompression
US5661822A (en) * 1993-03-30 1997-08-26 Klics, Ltd. Data compression and decompression
US5561724A (en) * 1993-06-11 1996-10-01 Hitachi Medical Corporation Method conducting smoothing processing in a linear direction having a minimized magnitude of density change
EP0679032A2 (en) * 1994-04-20 1995-10-25 Oki Electric Industry Co., Ltd. Image encoding and decoding method and apparatus using edge systhesis and inverse wavelet transform
US5598481A (en) * 1994-04-29 1997-01-28 Arch Development Corporation Computer-aided method for image feature analysis and diagnosis in mammography
US5481269A (en) * 1994-05-27 1996-01-02 Westinghouse Electric Corp. General frame wavelet classifier
US5712927A (en) * 1994-07-14 1998-01-27 Samsung Electronics Co., Ltd. Method and apparatus for binary-encoding image data using error diffusion with edge enhancement
US5761341A (en) * 1994-10-28 1998-06-02 Oki Electric Industry Co., Ltd. Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
P.J. Burt et al., "The Laplacian Pyramid as a Compact Image Code", IEEE Transaction on Communications, vol. No. 4. (Apr. 1983) pp. 532-540.
P.J. Burt et al., The Laplacian Pyramid as a Compact Image Code , IEEE Transaction on Communications, vol. No. 4. (Apr. 1983) pp. 532 540. *
R. Mont e far Chaveznava et al., Astronomical Image Coding using an Edge and Texture Model , Proceedings of the IEEE SP International Symposium on Time Frequency and Time Scale Analysis (25 28 Oct. 1994) pp. 148 151. *
R. Montefar-Chaveznava et al., "Astronomical Image Coding using an Edge and Texture Model", Proceedings of the IEEE-SP International Symposium on Time-Frequency and Time Scale Analysis (25-28 Oct. 1994) pp. 148-151.
Stephane Mallet et al., "Compact Image Coding from Edges with Wavelets", ICASSP 91: 1991 International Conference on Acoustics, Speech and Signal Processing, Toronto, Ontario, Canada (14-17 May 1991) pp. 2745-2748.
Stephane Mallet et al., Compact Image Coding from Edges with Wavelets , ICASSP 91: 1991 International Conference on Acoustics, Speech and Signal Processing, Toronto, Ontario, Canada (14 17 May 1991) pp. 2745 2748. *

Cited By (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6101277A (en) * 1994-10-28 2000-08-08 Oki Electric Industry Co., Ltd. Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US6148115A (en) * 1996-11-08 2000-11-14 Sony Corporation Image processing apparatus and image processing method
US6018368A (en) * 1997-07-11 2000-01-25 Samsung Electro-Mechanics Co., Ltd. Scalable encoding apparatus and method with improved function of scaling motion vector
US6898319B1 (en) * 1998-09-11 2005-05-24 Intel Corporation Method and system for video frame enhancement using edge detection
US6820199B2 (en) 1998-11-09 2004-11-16 First Data Corporation Sending electronic transaction message, digital signature derived therefrom, and sender identity information in AADS system
US6981154B2 (en) 1998-11-09 2005-12-27 First Data Corporation Account authority digital signature (AADS) accounts
US20020032860A1 (en) * 1998-11-09 2002-03-14 Wheeler Anne Mcafee Account authority digital signature
US20020112160A2 (en) * 1998-11-09 2002-08-15 First Data Dorporation Three Party Account Authority Digital Signature (AADS) System
US20020116608A1 (en) * 1998-11-09 2002-08-22 Wheeler Henry Lynn Sending electronic transaction message, digital signature derived therefrom, and sender identity information in AADS system
US20020129248A1 (en) * 1998-11-09 2002-09-12 Wheeler Lynn Henry Account-based digital signature (ABDS) system
US20020026575A1 (en) * 1998-11-09 2002-02-28 Wheeler Lynn Henry Account-based digital signature (ABDS) system
US6820202B1 (en) 1998-11-09 2004-11-16 First Data Corporation Account authority digital signature (AADS) system
US7549050B2 (en) 1998-11-09 2009-06-16 First Data Corporation Sending electronic transaction message for entity information account, digital signature derived therefrom, and sender identity information in AADS system
US20050005123A1 (en) * 1998-11-09 2005-01-06 First Data Corporation Account authority digital signature (aads) system
US7519821B2 (en) 1998-11-09 2009-04-14 First Data Corporation Account authority digital signature (AADS) system
US20030014372A1 (en) * 2000-08-04 2003-01-16 Wheeler Lynn Henry Trusted authentication digital signature (tads) system
US20020016913A1 (en) * 2000-08-04 2002-02-07 Wheeler Lynn Henry Modifying message data and generating random number digital signature within computer chip
US20020023217A1 (en) * 2000-08-04 2002-02-21 Wheeler Lynn Henry Manufacturing unique devices that generate digital signatures
US20090158029A1 (en) * 2000-08-04 2009-06-18 First Data Corporation Manufacturing unique devices that generate digital signatures
US7784106B2 (en) 2000-08-04 2010-08-24 First Data Corporation Manufacturing unique devices that generate digital signatures
US20130307929A1 (en) * 2011-02-18 2013-11-21 Sony Corporation Image processing device and image processing method
US9661301B2 (en) * 2011-02-18 2017-05-23 Sony Corporation Image processing device and image processing method
US10055814B2 (en) 2011-02-18 2018-08-21 Sony Corporation Image processing device and image processing method
WO2018005845A1 (en) * 2016-06-30 2018-01-04 Sony Interactive Entertainment Inc. Encoding/decoding digital frames by down-sampling/up-sampling with enhancement information
US10616583B2 (en) 2016-06-30 2020-04-07 Sony Interactive Entertainment Inc. Encoding/decoding digital frames by down-sampling/up-sampling with enhancement information

Also Published As

Publication number Publication date
KR960016579A (en) 1996-05-22
US5949910A (en) 1999-09-07
DE69525127T2 (en) 2002-10-02
US6101277A (en) 2000-08-08
DE69525127D1 (en) 2002-03-14
US5761341A (en) 1998-06-02
KR100388377B1 (en) 2003-11-28
EP0709809B1 (en) 2002-01-23
US5991448A (en) 1999-11-23
EP0709809A3 (en) 1997-02-05
EP0709809A2 (en) 1996-05-01

Similar Documents

Publication Publication Date Title
US5878172A (en) Image encoding and decoding method and apparatus using edge synthesis and inverse wavelet transform
US6330367B2 (en) Image encoding and decoding using separate hierarchical encoding and decoding of low frequency images and high frequency edge images
Strobach Tree-structured scene adaptive coder
US5974181A (en) Data compression system, method, and apparatus
US5838377A (en) Video compressed circuit using recursive wavelet filtering
JP3385077B2 (en) Motion vector detection device
US6389176B1 (en) System, method and medium for increasing compression of an image while minimizing image degradation
EP0866618A2 (en) Adaptive filtering for use with data compression and signal reconstruction
US5984514A (en) Method and apparatus for using minimal and optimal amount of SRAM delay line storage in the calculation of an X Y separable mallat wavelet transform
EP0491779A1 (en) Improved image compression method and apparatus
CN105408935B (en) Up-sampling and signal enhancing
EP1555833B1 (en) Method and apparatus for motion vector field encoding
US6591017B1 (en) Wavelet transform method and apparatus
US6768817B1 (en) Fast and efficient computation of cubic-spline interpolation for data compression
US7630568B2 (en) System and method for low-resolution signal rendering from a hierarchical transform representation
US7050641B1 (en) Method of providing extension to a signal and apparatus therefor
MXPA01010815A (en) Method and apparatus of creating and implementing wavelet filters in a digital system.
JPH09502840A (en) Encoder / Decoder for Television Image Subband Compatible Coding and Its Application to Hierarchical Motion Coding with Tree Structure
Young et al. Video compression using lapped transforms for motion estimation/compensation and coding
Sunkara Image compression using hand designed and Lifting Based Wavelet Transforms
KR101049638B1 (en) Image coding apparatus, image decoding apparatus, image band splitting apparatus and image band synthesizing apparatus
Maly et al. Fast lifting wavelet transform and its implementation in Java
Khalifa Review of Wavelet Theory and Its Application Toimage Data Compression
Fahmy et al. A Fast B_Spline Based Algorithm for Image Zooming and Compression
Boulgouris et al. Directional interpolation pyramids for still image compression

Legal Events

Date Code Title Description
STCF Information on status: patent grant

Free format text: PATENTED CASE

FEPP Fee payment procedure

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FPAY Fee payment

Year of fee payment: 4

FPAY Fee payment

Year of fee payment: 8

FPAY Fee payment

Year of fee payment: 12

AS Assignment

Owner name: OKI ELECTRIC INDUSTRY CO., LTD., JAPAN

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GO, SHIYU;REEL/FRAME:033501/0542

Effective date: 19951002

AS Assignment

Owner name: GLOBAL D, LLC., JAPAN

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:OKI ELECTRIC INDUSTRY CO., LTD.;REEL/FRAME:033546/0400

Effective date: 20140724

AS Assignment

Owner name: INPHI CORPORATION, CALIFORNIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GLOBAL D, LLC.;REEL/FRAME:034193/0116

Effective date: 20140729

FEPP Fee payment procedure

Free format text: PAYER NUMBER DE-ASSIGNED (ORIGINAL EVENT CODE: RMPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY